Chiller Design and Application New Technology 11/11/2015
Byron Hamm I received a graduate degree in Mechanical Engineering from Iowa State University with a specialty in the aerodynamic design of turbomachinery. I worked for over 40 years for a major manufacturer of centrifugal water chillers, having a part in the design of every machine currently produced. After retirement from that company, I joined GEA-consulting and have worked with Midea for more than four years as an engineering consultant. I continue to find this work fascinating, and have found my consulting work to be a continuing education. 2
Content Chiller system Compressor design Variable frequency drive Falling film evaporator technology Future trends in compressor technology 3
Chiller Systems External and internal considerations 4
A typical water chiller consists of - Evaporator to provide cooling water to the air conditioning system Compressor to provide system pressure differential and refrigerant flow Condenser to reject the cooling energy and motor power to the external environment Throttle device to reduce refrigerant pressure from condenser pressure to evaporator pressure This is shown schematically on the following slide 5
Typical simple chiller configuration - Chiller performance is determined from simple external measurements 6
Chiller system cycle diagram - Saturation line Region of subcooled liquid Region of superheated vapor Saturated liquid line Region of liquid and vapor mixture Saturated vapor line 7
Chiller system cycle diagram - Condenser process Throttle process Compression Evaporator process COP = Q evap / KW input Chiller performance is linked to cycle state points and component efficiencies 8
How can we improve the chiller performance? The most obvious method is to increase compressor efficiency and decrease mechanical losses. Compressor design always strives for this An additional method is to introduce subcooling into the condenser This increases the evaporator Δ H, decreasing the required mass flow for a given capacity, and thus reducing the amount of power required by the compressor This is shown on the following slide 9
Performance enhanced by using condenser subcooling - Subcooling Liquid saturation line Increases ΔH evap and decreases m evap decreasing compressor Wx and increasing COP 10
This is easy to accomplish in theory, but more difficult to accomplish in actual practice. In order to achieve subcooling, the subcooler tubes must be contained in a flooded compartment without a liquid/vapor interface. Controlling the liquid level to achieve this at part load is a design challenge. Midea uses an electronic flow control valve which senses condenser liquid level and regulates the refrigerant flow to ensure that subcooling is achieved as designed at all load conditions. 11
Another method to enhance chiller performance is to divide the compression process into two parts and use an economizer cycle as shown on the following slides. This enhances chiller performance in two ways: The evaporator ΔH is increased even more than with subcooling. Injection of the economizer vapor into the compression process slightly reduces the isentropic compression work input. This cycle also reduces the amount of discharge superheat that the condenser must remove. An economizer is in essence nothing more than a simple vessel to separate the liquid from the vapor at an intermediate pressure. 12
Chiller configuration with economizer - Pressure reduction is done in two steps Vapor is separated and injected into the compresso r 13
Chiller system cycle diagram with economizer - Economizer separates liquid and vapor Condenser process Vapor injection intercools the compression process Compression Evaporator process 14
Chiller system cycle diagram with economizer - Increases ΔH evap and decreases m evap decreasing compressor Wx and increasing COP Intercooling improves compression efficiency 15
Of course, subcooling can be included into the economizer cycle. This will alter the ratio of vapor to liquid that is provided by the economizer. A cycle code is used to evaluate the effects of different amounts of subcooling and the intermediate pressure level of the economizer. All Midea two-stage chillers incorporate the economizer cycle with subcooling for maximum chiller COP. These also incorporate electronic flow control valves to ensure that the liquid refrigerant is distributed to the various parts of the system according to the design intent. This is especially important at part load conditions. 16
Compressor design 17
Overall approach Design for required performance Cost reduce the design 18
Design for required performance The overall approach is to lay out the compressor flow path using a meanline design/analysis code, then examine the details using CFD. If the CFD analysis finds a problem that requires the redesign of a component, then we go back to the meanline analysis and examine the entire flow path to be sure that all components are still matched. The aerodynamic flow path design is done side-by-side with mechanical design to be sure there are no rotordynamic or other mechanical issues, and with review by manufacturing engineers to be sure the design is compatible with the available manufacturing capability. 19
Determine compressor basic flow path - Centrifugal compressors are dynamic compression devices they add energy by increasing flow velocity with the impeller and then diffusing this velocity to raise pressure. There are no enclosed volumes as in a positive displacement compressor (reciprocating, scroll, screw, vane, etc.) The governing equation for work input is the Euler equation Wx = (U 2 V θ 2 U 1 V θ 1 ) where U is the rotor velocity, V θ is the flow tangential (swirl) velocity, subscript 1 is the rotor inlet and subscript 2 is the rotor exit 20
Making use of the inlet and exit velocity triangles, this equation may be rewritten as Wx = ½( (V 2 2 - V 12 ) + (U 2 2 - U 12 ) + (V 1 2 - V 22 ) ) Change in absolute velocity Relative diffusion Change in radius For a given compressor design, work input is distributed among these components by the choice of impeller inlet and exit blade angles, rotational speed, inlet relative Mach number, and internal flow stability considerations. Different designers will choose different values for these components of energy addition according to what they feel is best. 21
Midea uses a meanline design/analysis code to evaluate the best combinations of these variables according to our best engineering judgement. Regardless of the detailed design choices that are made, it is clear that work input and flow velocity are closely related. For a single-stage compressor, the flow velocities are the result of adding the entire compressor work input in a single impeller. Flow velocities are high and the inevitable result is compressor noise. For a two-stage compressor, the work input for each impeller is only one half of the total, so the flow velocities are correspondingly lower. Two-stage compressors are much less noisy than single-stage compressors for this reason. 22
Typical two-stage compressor flow path - Volute/conic diffuser Crossover bend Diffuser diameter Vaned/vaneless diffuser Impeller diameter Vaned return channel Diffuser diameter Vaned/vaneless diffuser Impeller diameter First stage impeller Second stage inlet duct Second stage impeller 2
Important design decisions Impeller inlet relative Mach number/specific speed? Impeller inlet and exit blade angles? Impeller relative diffusion/d-factor? Stage work input split (impeller diameter ratio)? Vaned or vaneless diffusers? Pinched diffusers? Variable geometry vaneless diffuser? Diffuser radius ratio (and first/second difference)? Crossover bend diffusion? Start/end locations of vanes in vaned return channel? Single-vane vs two-vane return channel (second stage IGVs?)? Configuration of second stage inlet duct (conservation of rv θ ) Volute/conic diffuser geometry? 24
In addition to the aerodynamic factors, there may be geometric constraints as well Impeller hub diameter (power transmission, shaft critical speed) Impeller weight/axial spacing (shaft critical speed) Bearings and lubricant feed/drain/seal (shaft critical speed) Maximum casting diameters due to machine tool limitations Minimum volute location due to motor end-winding diameter Second stage IGV linkage (if present) Thrust balance piston and venting (if present) Overall chiller dimension envelope (shipping/application consideration) Material size/weight limitations (material handling within workshop) 25
The meanline design/analysis code is used to systematically study this array of variables in order to determine the best overall compressor design within the specified aerodynamic and mechanical constraints. This may include offdesign analysis to estimate range-to-surge, part load performance, etc. This process establishes the basic layout of the compressor. Only then are detailed CFD studies made of the individual components and component assemblies. If the CFD analysis indicates that some component must be redesigned, then the meanline design/analysis code is used to rebalance the entire flowpath to correctly accommodate the redesigned component. 26
The meanline design/analysis code evaluates many flow details 27
For some refrigerants, the isentropic compression process extends into the wet region. A very efficient compression may also extend into the wet region. Assuming dry compression will introduce fluid property errors. The meanline design/analysis code checks every state point and correctly calculates the refrigerant properties - Saturation line Compression into wet region 28
The meanline design/analysis code calculates impeller disk friction and seal leakage - The fluid in the cavity rotates at some fraction of impeller speed This creates a static pressure field Thus, the seal upstream pressure is known for seal leakage calculation The disk friction ΔH Is added to the leakage flow and mixed with the main flow The disk friction work input is also added to the shaft input power The recirculating leakage flow is added to the main flow to calculate impeller dimensions 29
The meanline design/analysis code calculates impeller axial thrust - The fluid in the cavities rotates at some fraction of impeller speed This creates static pressure fields Inlet flow has axial momentum mv Pressures are integrated over areas, and all forces are summed to calculate axial thrust 30
Impeller exit flow calculation uses the jet-wake model proposed by Japikse Region of high-energy isentropic core flow (jet) Region of low-energy boundary layer flow that accumulates in the suction surface corner (wake) This flow mixes to form the impeller exit flow. CFD analysis will give additional insight into the flow within the impeller 31
The blade profiles used in the vaned diffusers and vaned return channel can be either NACA 65-series or Double Circular Arc - NACA 65-series blade Double circular arc blade Either blade profile type may be specified, as well as the thickness/chord ratio, setting angle, camber, incidence, etc. Performance calculation is based on the extensive cascade data and calculation method presented in NASA SP-36, corrected from axial cascade to radial cascade 32
The meanline design/analysis code has many available options for cooling the hermetic motor - 33
These choices may each be analyzed for impact on cycle performance. A common and simple configuration is motor cooling using liquid from the condenser with vapor return to the evaporator. This is shown schematically below - Liquid from the condenser is fed into the motor cooling system Vapor is returned to the evaporator 34
This motor cooling scheme is simple. But note the following details This is a flow of refrigerant that is separate from and in addition to the flow required by the evaporator and condenser for the cooling load. Therefore, the compressor must provide this additional flow, and the compressor flow passage design must be altered accordingly. The condenser must have additional surface to condense this additional flow. The meanline design/analysis code automatically handles all mass flow requirements for any selected motor cooling scheme and incorporates these into the design of the compressor components. In effect, the compressor design is scaled up for both internal seal leakage flows and motor cooling flows. 35
Examples of the interaction of aerodynamic design with other design requirements 36
Example 1: The second row of vanes in a vaned return channel are made moveable to serve as second stage IGVs. They bind during operation, preventing rotation - Second vane in return channel is moveable to act as second stage IGV 37
Problem: In real life, moving parts require clearances. However, operating pressures cause deflection of parts. In this case, during operation the vanes bound against the castings and they could not be moved as designed. Solution: The first solution was to increase clearance on either side of the vanes. However, this caused unacceptable performance loss. The second solution was to strengthen the vanes with a steel insert inside the shaft (the vanes are an aluminum casting), and to strengthen the vane bearing structure and bolt it to a major structural casting. This solved the problem and actually improved performance because the vane clearance could be further reduced. The design team should include aero, mechanical, manufacturing engineers. 38
Example 2: FEA indicates impeller stress exceeds yield strength during the impeller overspeed test - Stress in impeller exceeds yield strength during overspeed test 39
Problem: FEA analysis indicates that the impellers will have areas of stress exceeding the yield strength of the material during overspeed tests. Various small changes to the design were not effective in reducing this stress (increased blade thickness, increased fillet radius, etc.). That is because, in hindsight, the root problem was the blade design in the inlet area Solution: The impeller vendor suggested a small change in the heat treatment of the impeller to increase the yield strength significantly. Tests showed that this change did not harm the machinability of the material. This change solved the problem without increasing the cost Designs should be done in collaboration with suppliers 40
Example 3: FEA indicates impeller deflection at operating speed - Deflection of impeller at operating speed 41
Problem: FEA analysis indicates that the impellers will deflect at both the tip and the inlet shroud area at operating speed. This may cause misalignment with the diffuser passage, interference with the surrounding castings, and interference at the inlet seal Solution: Location of the impeller should be done so that the deflected tip will align with the diffuser. Allow clearance around the impeller to avoid interference. Adjust seal clearance to avoid seal interference Component design and the design of surrounding components should consider operating deflections due to rotation, temperature, pressure, etc. 42
Example 4: CFD analysis indicates that the flow angle exiting the vaned return channel does not agree with the meanline prediction. Conservation of rv θ means the impeller leading edge will not be aligned with the flow as designed - Conservation of rv θ magnifies errors in flow angle leaving the return channel second blade 43
CFD solution for a tandem vane set in the return channel - 44
Problem: CFD analysis indicates that the flow leaving the return channel blades will have an exit angle 2-3⁰ different than indicated by the meanline analysis. The inlet duct is an area of conservation of angular momentum (constant rv θ ). Since there is a significant change in radius between the exit of the return channel blades and the impeller inlet, this error is magnified to be about 6-8⁰. The blades of the second stage impeller were designed to match the flow angles indicated by the meanline analysis. Solution: It was decided that the CFD analysis was probably more accurate than the meanline analysis in this area. The design of the second blade of the return channel was redesigned to provide the flow angle required by the impeller design. Various design methods should be compared where practical. 45
Cost reduce the design Consideration should be given at the design stage for ease of component manufacturing, ease of proper assembly, ease of service, and system reliability. The number of major castings should be as small as possible consistent with manufacture and assembly, in order to reduce potential leakage joints. External piping for the lubrication and motor cooling systems should have minimum connection points, and be designed for easy service while remaining leak tight. Carefully review the lubrication system, the heat exchanger designs, and the control system for function and cost. 46
Variable Frequency Drive 47
Variable frequency drive greatly enhances part load performance - Variable frequency drive is used to control capacity by adjusting compressor speed. Head is changed correspondingly. The typical chiller load line has decreasing head as capacity is reduced, so this works well. Variable frequency drive can usually be used down to about 50% of design speed before encountering surge (AHRI part load conditions). Below that, control reverts to IGV but requires much less vane closure due to already decreased capacity from speed reduction. The plot on the next slide shows a comparison of COP for a chiller using VFD down to 50% load and then IGV control at 25% load, compared to the same chiller using only IGV capacity control. The test points are those specified by AHRI for the calculation of IPLV (Integrated Part Load Value). 48
Variable frequency drive greatly enhances part load performance - IGV capacity control VFD capacity control Calculated IPLV 6.602 10.883 49
Variable frequency drive greatly enhances part load performance There is a small penalty at full load due to some efficiency loss in the VFD drive. There is a large performance gain in the mid-range, the test points that weigh most heavily in the calculation of IPLV. The IGV control at 25% load still produces better efficiency because the guide vanes do not have to be closed very much. VFD control is optional on Midea gear-drive compressors. VFD control is required on all high-speed direct-drive compressors in order to achieve the required compressor speeds. 50
Falling Film Evaporator Technology 51
What is a falling film evaporator, and how is it different from the traditional flooded evaporator? 52
Flooded evaporator Dry vapor to compressor Tube bundle Filter to remove liquid droplets Liquid distributor Liquid inlet 53
Flooded evaporator Dry vapor to compressor Large vapor space so low velocity flow will not carry liquid droplets into compressor Tube bundle extends to shell to prevent gas bypass Liquid level in shell, optimized so that boiling action just wets the top tubes for maximum performance Flow is distributed evenly along the length of the shell Liquid inlet 54
Falling film evaporator Baffle to remove liquid droplets Vapor path for generated vapor Tube bundle Dry vapor to compressor Liquid distributor above top row of tubes Lower tube bundle to evaporator any remaining liquid Liquid inlet 55
Falling film evaporator Baffle and vapor space around distributor designed to minimize carryover of liquid droplets to compressor Dry vapor to compressor Vapor must exit the tube bundle and flow around the tube bundle Liquid level in shell Liquid inlet 56
How well does this design concept work? 57
How well does this design concept work? This test data shows that, if a proper film is developed, all the tubes in the column will have equal heat transfer coefficient. Above a minimum flow rate, the heat transfer coefficient is very constant 58
How well does this design concept work? This test data illustrates that the heat transfer coefficient can be higher in a falling film configuration compared to a flooded configuration 59
How well does this design concept work? This test data illustrates that different enhanced tube surfaces will have different performance characteristics 60
Potential benefits: Reduction in working fluid to about 1/3 of flooded evaporator Better heat transfer performance More uniform heat transfer throughout the evaporator Lower approach temperature More compact evaporator design Superior for oil management (oil holds up in a flooded bundle but drains to the bottom of a falling film evaporator) 61
Potential disadvantages: Less design experience with falling film evaporators Uniformity of liquid distribution onto the top tube row is a design challenge Less tolerance for undercharging of the chiller There is more to performance/cost than boiling heat transfer coefficient Does the performance of a falling film evaporator reduce the cost of the tube surface and refrigerant charge enough to offset the added cost of the distributor and flow baffles? 62
Charge optimization Charge optimization is the same for either evaporator type find the minimum refrigerant level/flow rate that fully utilizes the heat transfer surface ITD/ΔT or Approach Optimum charge is where the curve flattens out Refrigerant charge ITD = inlet water temperature saturation temperature ΔT = inlet water temperature outlet water temperature Approach = outlet water temperature saturation temperature 63
Future Trends in Compressor Technology 64
The compressor world is changing: Gen 1 Gen 2 Traditional design High-speed gear-drive Compact size Efficiency + IPLV Gen 3 High-speed direct-drive + VFD (no gears, fewer bearings, permanent magnet motors) (variable speed) 65
Increasing compressor speed drives motor development: Gen 1 Gen 2 Gen 3 Traditional induction motor High-speed Induction motor Permanent Magnet motor Compact size Efficiency 66
Increasing compressor speed drives bearing development: Oil-lubricated sleeve bearings Oil-lubricated ball bearings Oil-free Magnetic bearings Ceramic ball bearings Gas bearings Foil bearings 67
Bearing development will result in new requirements: New manufacturing methods Aerodynamic performance of the compressor due to seal clearance Stiffness and rotordynamic characteristics System requirements Continuing changes to satisfy new requirements of size, cost, efficiency, and oil-free operation In the following slides we will briefly review the pros and cons of several bearing types 68
Oil-lubricated sleeve bearings: Pro Traditional design Simple, reliable, and relatively inexpensive Medium clearance requires medium seal clearance and leakage Medium load capability Con Requires lubrication system oil pump, filter, heater, cooler, pressure regulator, piping Requires oil reclaim system Speed limited by increasing bearing losses Misalignment issues may require more expensive pivoted-shoe design Considered the baseline design for this comparison Still perhaps the best solution for the largest lower speed compressors 69
Oil-lubricated ball bearings: Pro Well known design methods and operating characteristics Low clearance results in low seal clearance and leakage Low bearing friction Allows higher compressor speed, limited by internal bearing loads High load capability, compact bearing system Con Requires lubrication system oil pump, filter, heater, cooler, pressure regulator, piping Requires oil reclaim system Higher cost than a sleeve bearing Lower loss alternative for all compressor sizes within the speed limitation of this bearing type 70
Magnetic bearings: Pro Oil free Known technology, relatively low project risk No bearing friction Can be programmed to eliminate balancing and critical speed issues No compressor speed limit Con Requires complex control system Requires backup bearing system and can be bulky High clearance results in high seal clearance and leakage Expensive Low-loss oil-free alternative for all compressor sizes and speeds, but high required clearance will compromise performance of the smaller compressors If we separate magnetic bearings from permanent magnet motors, they can be used for higher capacity compressors 71
Refrigerant lubricated ceramic ball bearings: Pro Oil free Low clearance results in low seal clearance and leakage Lower bearing friction than oil lubricated ball bearings Higher compressor speeds than oil lubricated ball bearings High load capability, compact bearing system No oil reclaim system Con Requires simplified lubrication system pump, filter, piping Newer technology requiring some project development Higher cost than an oil lubricated ball bearing Low-loss oil-free alternative for all compressor sizes within the speed limitation of this bearing type 72
Gas (air) bearings (hydrostatic): Pro Oil free Similar in appearance to the simple sleeve bearing, but uses gas pressure to lift the shaft, not a film generated by shaft rotation Very low clearance results in very low seal clearance and leakage Very low bearing friction No compressor speed limit Con Requires external gas source with controlled pressure and temperature Low load capacity requires larger bearings New technology requiring significant project development Very low clearance requires precision manufacture and assembly Cost unknown but likely to be higher than ordinary sleeve bearings Low-loss oil-free alternative for all compressor sizes with no speed limitation Low clearance would be a special benefit for small-medium compressors 73
Foil bearings (hydrodynamic gas bearings): Pro Oil free Simple concept that requires no lubrication system at all Higher clearance results in higher seal clearance and leakage Very low bearing friction No known compressor upper speed limit Con Requires high minimum compressor speed, thereby limiting IPLV Operating concept requires low loads and very high speeds New technology requiring significant project development Requires precision manufacture and assembly Cost unknown but likely to be very high Current state of technology is only for very small very high speed machines Requires extensive development for each application 74
Approximate range of application of bearing technologies: I believe the newer bearing technologies are best applied at the lower end of the capacity range for relatively small high-speed direct-drive compressors. 75
Midea has ongoing projects to evaluate each of these technologies 76
Two-stage Falling-film Centrifugal Chillers Chiller Plant System Control Water-cool Falling-film Screw Chillers Air-cool Falling-film Screw Heat Pump
78
Thank You! 79