Thermal Aspects of a Shipboard Integrated Electric Power System

Transcription

1 Thermal Aspects of a Shipboard Integrated Electric Power System Christopher R. Holsonback and Thomas M. Kiehne Abstract Development and validation of a dynamic model focused on thermal aspects of the integrated propulsion system (IPS) on a notional all-electric ship (AES) is presented. This model serves as a baseline for investigation of thermal management technologies and architectures focused on identifying tradeoffs, improving system-level efficiency, and increasing the performance of an AES. The IPS model includes component models for gas turbine engines, synchronous generators, motor converters, propulsion motors, fixed-pitch propellers, and a ship hull. These component models are integrated into a system-level simulation and dynamic results for a ship-maneuver called a crash-back are presented. The resulting thermal-mechanical dynamics are discussed in depth. A I. INTRODUCTION new paradigm in Navy ship building is electrification of major systems and the use of electricity as the primary medium of energy transfer throughout the ship. As with any paradigm shift, the transition to an all-electric ship (AES) faces significant technical and institutional challenges. However, due in part to relatively recent advances in high torque-density electromagnetic motors and high power semiconductor switches, future naval surface ships can be powered through an integrated propulsion and electrical power distribution system, often called an integrated power system (IPS). In this architecture, all prime movers onboard a ship are coupled to generators, resulting in a large quantity of electrical power that can then be distributed throughout the ship. The vast majority of ship systems are then powered electrically, eliminating many of the hydraulic, pneumatic, and mechanical systems on current ships. The inherent flexibility of an IPS will reduce the total number of prime movers required by the ship, now typically four gas turbines with a total installed power of 8-9 MW. An IPS will also allow for scheduling of the prime movers, distributing the electrical load in such a way as to optimize fuel consumption and dramatically increase the electrical power available for other systems, such as future electric weapons and advanced sensors [1]. The attributes of the AES are unique and have essentially no analogy in land-based systems. Implicit in the need for a robust electrical system is the need for an equally robust cooling system. In fact, the US Department of Defense has explicitly stated that heat is the second leading cause of the Manuscript received February 27, 21. This work was supported by the Office of Naval Research under the auspices of the Electric Ship Research and Development Consortium (ESRDC). C.R. Holsonback was a graduate research assistant with the University of Texas at Austin. He is now with General Electric Energy, Greenville, SC 29615, USA ( christopher.holsonback@ge.com). T.M. Kiehne is with the Applied Research Laboratories, the University of Texas at Austin, Austin, TX USA ( ; kiehne@arlut.utexas.edu). failure of electronic equipment on the battlefield [2]. The thermal management situation becomes even more challenging when one realizes that the cooling system must meet the needs of distributed kilowatt (kw) and megawatt (MW)-sized heat loads, including large electric generators and propulsion motors, radar and sonar systems, pulsed electric weapons, and numerous power electronic converter cabinets. A dynamically reconfigurable and controllable power system requires a reconfigurable and controllable thermal management system. Work reported here addresses a dynamic, physics-based, thermal-mechanical model of a shipboard IPS which might serve as a baseline for characterization of advanced shipboard thermal management technologies and architectures. The model is implemented in ProTRAX, a dynamic software tool designed for commercial, land-based power plants. Custom coding within ProTRAX, and other modeling environments, has been used to develop models of IPS components peculiar to the US Navy, which have been individually validated against data from the open literature. II. IPS THERMAL LOADS While a ship floats in a nearly infinite heat sink, removal of waste heat from equipment within a ship represents a significant challenge. In a separate paper [3], the authors have attempted to quantify the thermal loads expected on a notional AES. A brief summary of major findings in that paper follows. The assumption is made that the ship carries two 36 MW main gas turbines (MGT) and two 4.5 MW auxiliary gas turbines (AGT), modeled after the Rolls-Royce MT3 and RR45, respectively. Taking genset scheduling, variable motor and drive efficiency, and distribution efficiency into account gives the genset output power plot of Figure 1. At low ship speeds, one AGT serves to meet the ship service load while the other allows for propulsion. Thereafter, one MGT takes over, providing both ship service and propulsion loads. When this MGT reaches its rated power, an AGT serves as the bridge until power demand warrants the second MGT. Above this, all four gensets are required. The second MGT is not used until ship speed exceeds approximately 27 kt. Figure 2 indicates that a maximum waste heat load of approximately 11 MW is produced at full speed. If this load were actively cooled by onboard chillers, more than 3,1 tons of refrigeration would be required. Fortunately, the propulsion motors, converters, and generators are typically seawater cooled. If only ship service loads and distribution losses are actively chilled, approximately 1, tons of refrigeration will be needed at full speed. Depending on the coefficient of performance of the chillers, this amount of chilling may require between MWe of dedicated

2 power. Incorporating less efficient ship service loads (e.g., advanced radar, free-energy lasers, rail guns, etc.) will increase this heat load and power requirement dramatically. If for example the motor converters were actively cooled for increased reliability, the thermal load would increase to approximately 1,8 tons of refrigeration at full speed. Depending on the performance of the chillers, this amount of chilling may require MWe of dedicated power. As seen in Figure 3, a significant amount of energy remains in the gas turbine exhaust, especially at high power levels. At full speed nearly 13 MJ of thermal energy from the fuel is exhausted to the atmosphere every second. While cruising at 14 kt, more than 3 MW is rejected to the environment. Fuel usage increases in lock step with the exhaust heat. (HPT, IPT, and FPT). Using these parameters, a steady-state model exhibited an exhaust temperature of 465ºC, very close to the quoted value of 466ºC [4]. Under these conditions, the combustor outlet temperature was found to be 1177ºC. Total Genset Power [MW] MT3 #2 MT3 #1 RR45 #2 RR45 #1 Fig. 3. Turbine exhaust heat and total fuel flow rate versus speed Ship [kn] Fig. 4. Representation of MT3 with station designations. Fig. 1. Total genset output power taking scheduling into account. 12 Motors & Converters 1 Distribution Waste Heat [MW] 8 Generators Ship Service The AGT model is patterned after a RR45 with a nominal output of 4.5 MW. The assumed physical layout of the RR45 is shown in Figure 5. The compressor overall pressure ratio was set at 13.5:1 and the optimization process resulted in GGC, GGT, and FPT (see Figure 5) isentropic efficiencies of approximately 86%. The steady-state model produced a heat rate of 1,84 kj/kw-hr and the exhaust temperature was found to be 518ºC, matching published values very well. The combustor outlet temperature was found to be 171ºC (1344 K), which is high for this turbine Ship [kn] Fig. 2. System waste heat loads as a function of ship speed. III. IPS COMPONENT MODELING AND VALIDATION A. Gas Turbine Modeling The MGT model is based on a 36 MW Rolls-Royce MT3 engine shown in Figure 4; it has a high pressure compressor (HPC) ratio of 4.2 and an intermediate pressure compressor (IPC) ratio of An optimization process resulted in isentropic efficiencies of about 86% for the IPC and HPC and 87% for the high, intermediate, and power turbines Fig. 5. Representation of the RR45. The steady-state model served as the basis for a dynamic model created in the commercial, thermodynamic-cycle analysis software ProTRAX [5]. ProTRAX was designed for dynamic simulations of thermodynamic cycles found in utility power plants. Applications include power plants and industrial power complexes. ProTRAX is fully dynamic and

3 models exist for a wide range of system elements including gas turbines, heat exchangers, auxiliary cooling components, pipes, inter-coolers, regenerative components, electrical components, and system-level controls. The software has traditionally been used to examine plant startup, shutdown, and/or emergency operation. Dynamics play a role in the response of an IPS to changing conditions, such as increased or decreased power demand. Models specific to the system represented here were created within ProTRAX. Where Navy peculiar models did not exist, these were also created. Results shown in Table I indicate that the dynamic MGT model reproduced design-point performance at ISO conditions in nearly every measurable parameter. For non- ISO conditions #1 and #2, the fuel flow rate, SFC, exhaust temperature, and combustor outlet temperature generally increase with increasing ambient temperature, inlet losses, and exhaust losses. Conversely, the inlet air flow rate and thermal efficiency generally decrease with increases in these parameters. Note also that the thermal efficiency is higher and SFC is lower at 4 MW output than at rated conditions, implying that this is a preferred operating condition. TABLE I PROTRAX MGT MODEL PERFORMANCE. SFC [kg/kw-hr] HPC Bleed Flow Rate [kg/s] MT3 SFC vs. Brake Power Syntek MT3 model SFC error % %.38 1.% %.28 5.% %.18.% Brake Power [MW] Fig. 6. Validation of ProTRAX MGT model at off-design conditions. MT3 Bleed Schedule Bleed Flow Bleed Percent % 4% 35% 3% 25% 2% 15% 1% Relative Error [%] Bleed as Percent of Intake Flow [%] 5% % Brake Power [MW] As indicated in Figure 6, the ProTRAX model also simulates response of the MGT engine well at off-design conditions. The data used for off-design conditions were a table of engine SFC versus generator output presented in [6] at non-iso condition #1. The model results agree with the SFC data within 5% down to loads as low as 18 MW. Experimentation with the model made it clear that specific fuel consumption (SFC) performance could be forced to follow available data by inserting a compressor bleed after the HPC that vented to ambient (more accurately described as a blow-off ). Since the air exiting the HPC has had considerable work done on it by the IPC and HPC, venting some portion increases the SFC at each operating point. A bleed-schedule that determines the amount of post-hpc air to vent at each load was determined manually through experimentation on the MGT model. It is presented in Figure 7. The schedule output serves as the set point for a PI controller that dynamically modulates a bleed valve during runtime. Utilizing this bleed schedule causes the MGT dynamic model to exhibit the SFC characteristics provided in [6] exactly, as illustrated in Figure 8. The model thermal efficiency is also shown in this figure. SFC [kg/kw-hr] Fig 7. MGT Dynamic Model HPC Bleed Schedule. MT3 Dynamic Model Performance Model SFC Syntek SFC Thermal Efficiency. % Brake Power [MW] Fig 8. MGT dynamic model SFC and thermal efficiency. The AGT dynamic mod was also implemented in el ProTRAX. As indicated in Table II at non-iso conditions, results of the AGT model matched available design-point performance from [7] extremely well. The model over predicts the exhaust flow rate by about 2% and under 45% 4% 35% 3% 25% 2% 15% 1% 5% Thermal Efficiency [%]

4 predicts the exhaust temperature by about 2.3%. These parameters are very sensitive to the brake power demand, which was estimated via the known generator output power and an assumed 96% generator gearbox efficiency. The resulting thermal efficiency and SFC are reasonable when compared to engines of similar rating. The model predicts a combustor outlet temperature of about 11ºC, which satisfactorily reflects estimates from steady-state analysis. TABLE II PROTRAX AG T MODEL PERFORMANCE. Parameter Unit 51- K ISO non- ISO #1 non- ISO #2 genset output power MW estimated gen eff % brake power M W inlet temperature deg C relative humidity % inlet pressure loss in. H2O 4 4 exhaust pressure loss in. H2O 8 8 fuel flow rate kg/s inlet air flow rate kg/s exhaust flow rate kg/s exhaust temperature deg C firing temperature deg C SFC k g/kwhr thermal efficiency unitless 32.8% 3.7% 28.3% NOTE: Ambient pressure = bar, fuel LHV = 47, 12 kj/kg To develop the SFC curve for the AGT, genset output power was varied from 5.3 to.5 MWe. Results are depicted in Figure 9 and compared with results in [8]. Performance of the ProTRAX model matched the SFC model within 1% relative error at all operating points. Deviation between expected and simulated SFC is due to a combination of factors such as lack of 1) compressor bleed, 2) auxiliary power loads, and 3) variable compressor geometry. B. Propulsion Motor Modeling In 1995, the Navy commissioned a large electromagnetic motor capable of providing ship propulsion for a future electric ship. This came to be called the Advanced Induction Motor (AIM) because of characteristics such as high power factor and efficiency, large air-gap for shock standards, and variable-speed capability. Many of the desirable features of the AIM resulted from it being powered through a pulsewidth modulated (PWM) converter, which traced its evolution to decades of advances in high voltage and high current semiconductor switches [9]. Reference [9] indicates that the IPS AIM was designed for 19 MW at 15 rpm. At rated conditions, it has an efficiency of 95.7% and a slip of 1.23%. Fifteen phases, arranged to create twelve poles, power the stator. The stator phases are arranged in three symmetrical groups of five, which allows the motor to run on five, ten, or fifteen phases as desired. Reference [1] establishes that the AIM is fundamentally an air-cooled machine, with top-mounted air to-water heat exchangers that are controlled to be used only when necessary. SFC [kg/kw-hr] RR45 SFC vs. Brake Power Davey PT RR45 model Error. % Brake Power [MW] Fig. 9. AGT model off-design performance validation. 1% Three AIM-based performance data points exist in the literature [11], [12]: 1) 19 MW at 15 rpm, 1.21 MN-m (rated condition), 2) MW at 127 rpm, MN-m, and 3) 17.3 MW at 12 rpm MN-m. Unfortunately, additional data such as slip, efficiency, and power factor are only known for the rated condition. Reference [13] provides target efficiency for the AIM at various speeds as a percentage of 18 rpm, which is presented in Table III. TABLE III AIM EFFICIENCY VERSUS SPEED [% of rated] Hard Target Efficiency [%] Stretch Target Efficiency [%] Reference [9] describes many of the characteristics of the motor converter. The converter utilizes PWM technology to control both the voltage and frequency of each phase supplying the motor stator. Using PWM allows for the use of an asynchronous induction motor with all of its associated advantages. PWM also introduces lower harmonic distortion into the ship electrical grid and motor, thus reducing noise and enabling more-effective, low-pass filtering. A practical, per-phase equivalent circuit, shown in Figure 1, is used in this analysis. Subscripts s, m, and r refer to stator, magnetizing, and rotor branches, respectively. The magnetizing inductance represents the portion of the magnetic field linking the rotor and stator. The rotor and stator inductances represent those portions of the rotor and stator magnetic fields that do not link (thus, they are leakage voltages). The rotor and stator resistances are a result of the non-ideal nature of rotor and stator windings. While use of a practical equivalent circuit of this motor significantly simplifies performance calculations for the IPS AIM, it also introduces the need to accurately determine the resistances and inductances of equivalent circuit elements. Parameters determined using equivalent circuit parameter (ECP) estimation are presented in Table IV. These parameters were determined by matching to available data in the literature and result in the IPS AIM motor model having a rated power of 21.5 MVA and a rated power factor of.92. 8% 6% 4% 2% Relative Error [%]

5 Fig. 1. Induction motor practical equivalent circuit. TABLE IV IPS AIM MOTOR MODEL ECP SET Parameter Unit Value R s Ω 2.76E-1 L s H 1.13E-2 L m H 4.4E-1 R r Ω 1.14E-1 L r H 5.8E-3 The first iteration of a dynamic IPS AIM model was developed in MATLAB and followed by a custom-coded model developed in ProTRAX. In this analysis, the slip speed was varied linearly with time. An illustration of the slip speed change during a crash-back simulation is shown in Figure 11. A crash-back is a highly dynamic, emergency maneuver whereby the ship is traveling ahead at full speed and demand is then changed to backing at full speed. Motor [rpm] Ship [knots] S m S g -S m Motoring at Rated Torque Reduced Torque Motoring Motoring Reduced Power Constant Power Regeneration Generating Reduced Torque Motoring Motoring time Motoring at Rated Torque time Fig 11. Illustration of motor and ship speed transition during a crash-back. In a first iteration of the ProTRAX IPS AIM model, only a single motor and controller were implemented, along with a single shaft, propeller, and ship modules. A PID controller was used to modulate controller demand from +/-1% rated shaft speed depending on the difference between the then current ship speed and a set point as determined by the operator. After initial testing of the AIM model was completed, the four pairs of motors and controllers were implemented in the ProTRAX IPS model, along with two shafts and two propellers. One motor controller serves as the master, receiving speed demand from the PI control system. It then passes this information to the other three controllers, which serve as slaves. An efficiency curve from [15] served to model the losses in the controller. Initial testing of the ProTRAX motor model indicated that the above refinements to the dynamic model increased the torque limit of the motor, limited regeneration to 1 MWe, and limited the acceleration, deceleration, and transition time of the motor. However, the response of the motors and converters to changes in speed demand differed significantly from the MATLAB model. Characteristics of the propeller and propeller-ship interactions result in a highly nonlinear shaft-speed versus ship-speed curve, especially during emergency maneuvers such as a crash-back. Because the response of the motor is so dependent on the characteristics of other elements of the model, integrated results for the ProTRAX version of the IPS AIM and motor converter models, as well as the generator and distribution system models, are presented following discussion of the shaft, propeller, and ship models. C. Shaft, Propeller, and Ship Modeling Quantification of gross mechanical-electrical interactions of the propellers and ship structure is critical to predicting operating conditions of the motors and gensets during ship transients. If not properly quantified and controlled, this tightly coupled system could become unstable. Characteristics of the modeled propeller [6] are given in Table V. Reference [6] also provides estimated hull powering data for a baseline AES. Knowledge of the powering data and propeller characteristics allows for determination of hull resistance. Since ProTRAX was designed for land-based, commercial power plants, it clearly does not have native models for the propeller, propeller shaft, and ship hydrodynamics. Consequently, these elements were developed from first principles and custom coded in ProTRAX. Two propulsion motors were assigned to power each shaft, and two shafts and propellers accelerate the ship. TABLE V CHARACTERISTICS OF PROPELLER Parameter Symbol Value Unit blade number Z 5 expanded area ratio *1 EAR 75 diameter D 23 ft pitch-to-diameter ratio P/D 1.4 weight W 1.1E5 lbf mass moment of inertia MOI 3.65E6 lbm-ft 2 entrained water MOI MOI ew 9.11E5 lbm-ft 2 Since the propeller shaft, propeller, and ship models are passive elements, they require power input from the motor models in order to function. However, response to changes in demanded ship speed is intricately connected to other components of the IPS system. As such, simulation results for these components are difficult to separate from the rest of the model, and will not be provided here. A discussion of the characteristics of the propeller and ship systems will be given in concert with results of the overall IPS model below.

6 D. Generator and Distribution System Modeling From a thermal management standpoint, generators and electrical distribution system are well understood. Therefore, little effort was expended to model these components. Consequently, the generator was modeled as a simple gyrator in which the current demand from the distribution system is converted to an input mechanical power demand. The ship service grid was assumed to be a constant load at 4 MWe. The efficiency of the generators was assumed to vary with the input power from the turbine. A characteristic curve from [15] was used for both of the generators, with the main and auxiliary generator efficiencies peaking at 98% and 95.5%, respectively. Again, simulation results for these components are difficult to separate from the rest of the model, and are not provided separately here. IV. SYSTEM-LEVEL SIMULATION Using ProTRAX, the component models described above were integrated to form a single dynamic model of the entire IPS. Information flow during a simulation is illustrated in Figure 12. This IPS model was used to simulate ship maneuvers and resultant gross mechanical-electrical and thermal effects on the IPS. Model response is best demonstrated via simulation of several highly dynamic events; thus, a full-ahead to full-astern crash-back and acceleration from full-stop to cruise were modeled. Due to space limitations, only the crash-back results are presented here, with a focus on thermal aspects of this maneuver. The crash-back begins at full-ahead (+1% or 29.4 kt) with a single motor mechanical full power of 17.2 MW (68.8 MW total) at rpm. The demand then transitions to full-astern (-1%), during which the motor mechanical power ramps down at a rate of approximately 2 MW/s with motor regeneration beginning at 1 MWe within 2 s of the start of the crash-back. The ship speed passes through zero at about 95 seconds into the crash-back. After this point, the motor power, motor speed, and ship speed increase quickly to their steady-state values. The ship ends the crash-back at kt with a motor mechanical power of 17.5 MW (69.9 MW total) at rpm. The power difference between the ahead and astern condition is caused by the characteristics of the propeller and the assumed Taylor wake fraction. Figure 13 shows the thrust and reaction torque created by the propeller during the crash-back. These are functions of the propeller advance angle relative to the quiescent water, which is also shown in the figure. The advance angle begins at about 25 degrees during full-ahead operation. It then transitions slowly to about 4 degrees during regeneration and passes through 9 degrees to enter the crash-back quadrant (9-18 o ) of operation at which the motors begin motoring astern. The crash-back quadrant lasts about 35 seconds, after which the advance angle exceeds 18 degrees and the ship begins moving astern. The propeller torque and thrust begin the crash-back at relatively high values, drop to slightly negative values during regeneration, become very negative during the crash-back operation, and settle into slightly lower (less negative) steady-state values. Note that reaction torque is relatively symmetric, while thrust is asymmetric, especially during the later portion of the crashback. Results for ship and motor speed, motor power, generator electrical power, and power turbine speed are all provided in the source document [16]. PI Shaft Controller Ship Service Loads Motor Converter PID Gas Turbine Frequency Current Demand Voltage Current Power Voltage Shaft Grid Propeller Torque Torque Thrust Prop Shaft Motor Demanded Power Voltage Current Current Generator Ship Ship Fig. 12. Variable flow throughout IPS model at runtime. Fig 13. Propeller advance angle, torque, and thrust during crash-back. Ship Hull Generators, motors, and converters are all significant sources of waste heat in the IPS. Figure 14 illustrates the evolution of these heat loads during the crash-back. At the beginning of the maneuver, main and auxiliary generators produce 724 kw and 2 kw of waste heat each (1.85 MW total). Each motor-converter pair produces 1.38 MW of waste heat (5.5 MW total). Therefore, a total of 7.35 MW of waste heat, not including gas turbine exhaust energy, is created by these components. During regeneration, the heat output of the motor-converters remains at a nominal 3-33 kw since they are producing electrical power. Chaotic, transitory heat output from each of these components occurs during the transition from regeneration to motoring astern. Hesitation of the propulsion motors during astern motoring

7 is clearly reflected in the waste heat of the motor-converter pair. At full-astern, main and auxiliary generators produce 729 kw and 23 kw of waste heat each (1.86 MW total). Also at full astern, the motor-converter pairs produce 1.41 MW of waste heat each (5.63 MW total), resulting in a total waste heat of 7.5 MW. Fig 15. Energy transfer via gas turbine exhaust during crash-back. Fig 14. Generator and motor-converter waste heat during crash-back. Figure 15 shows the energy transfer rate to ambient via the exhaust during the crash-back. The exhaust of each MGT transfers 53.6 MW of fuel energy to ambient prior to the start of the crash-back; the exhaust of each AGT transfers approximately 11 MW. The total energy lost via the exhaust is about 128 MW. During the regeneration phase of crash-back, both MGTs and AGT #2 are loaded with the idle load while AGT #1 supplies the small electrical power demand. Each continues to produce hot exhaust. However, the nominal exhaust energy transfer rate drops to approximately 5 MW for about 4 seconds. As electrical demand from the propulsion motors increases, it feeds back into the power system and slows the turbines, to which power turbine speed controllers respond by increasing fuel flow. Therefore, the energy transferred to ambient via the exhaust increases with increasing load. As with the generators, the effect of hesitation on the propulsion motors is evident in the gas turbine exhaust. At the end of the crashback, the total exhaust energy transfer rate to ambient is greater than 13 MW. Figure 16 shows the exhaust gas temperature from the main and auxiliary turbines during crash-back. The MGTs and AGT #2 begin the crash-back at an exhaust temperature of 46ºC and 532ºC, respectively. AGT #1 begins the crashback at an exhaust temperature of 62ºC due to its slightly lower spool speed and air mass flow rate. During motor regeneration, the MGT exhaust temperature remains remarkably constant, increasing only briefly to 75ºC resulting from the additional fuel injected by the speed controller to provide the power for astern motoring of the propulsion motors. The exhaust temperatures of the AGTs, on the other hand, drop significantly during regeneration. However, the exhaust temperatures of AGT #1 and #2 recover quickly, returning to their starting values after a brief excursion to 68ºC and 6ºC, respectively. Fig 16. Gas turbine exhaust temperature during crash-back. V. EVALUATION OF RESULTS In general, as summarized in Table VI, the agreement between ProTRAX simulation results and the initial systemlevel thermal calculations presented earlier, and in [3], is excellent. At similar shaft power levels, there is only a very small relative difference in the calculated total motorconverter waste heat and total fuel heat release remaining in the exhaust, denoted exhaust heat in the table. The largest relative difference, of less than 3%, occurs in calculation of the generator waste heat. Comparison can be made between the crash-back results and the simulations of a notional destroyer-class, all-electric ship presented in [17]. Integrated power system components presented therein are quite similar to those utilized in this analysis because both rely on [6] and knowledge of the IPS. However, the ship hydrodynamic and propeller models used in [17] are very different from those used in this analysis. Further, the propulsion motors modeled in that work were rated for 36.5 MW at 12 rpm, giving a rated torque of 2.9 MN-m. The rated combined torque of two AIMs used in this analysis is 2.74 MN-m. Nevertheless, the propulsion motor speed versus time plot is qualitatively similar between this work and [17], as shown in Figure 17.

8 TABLE VI COMPARISON OF DYNAMIC IPS MODEL RESULTS TO INITIAL ESTIMATES. Dynamic Initial Model Relative Parameter Unit Calculations Results Difference Ship kt % Shaft Power Motor- MW % Converter Heat MW % Generator Heat MW % Distribution Heat MW 1.6 NA NA Exhaust Heat MW % Finally, improving the efficiency and fuel consumption characteristics of the prime movers should be of paramount importance to the Navy given the significant effect fuel costs will have on the lifecycle costs of future ships. Increased use of electrical devices for propulsion and auxiliary systems will shift more of the system-level inefficiency onto the prime movers. Consequently, the efficiency characteristics of the prime movers will impose an upper limit on the system-level efficiency of future all-electric ships. This should provide the impetus for the Navy to explore means for efficiency improvement in their future ships to include exploration of unconventional architectures and power sources. REFERENCES Inset and result from Andrus et al. (26) Fig 17. Comparison of crash-back results to those of Andrus et al. (26). In Figure 17, the inset from [17] is for a ship crash-back from 3 kt, ~3 MW per shaft at 11.3 rpm to -5 kt, 3 MW per shaft at -75 rpm over a period of 8 seconds. Scaling the inset to match the scale of the background plot gives the image shown. Clearly, the same characteristic motor speed versus time behavior is exhibited by both IPS system models. However, because of the higher rated torque, slower initial rotational speed, and higher power ramp rate limit, the crash-back response from [17] is clearly faster than that of this analysis. In fact, [17] reported a time to stop of 63 seconds versus the ~95 seconds reported here. V. CONCLUSIONS In this work, excellent agreement was achieved between initial waste heat estimates and dynamic model results. This provides a high-level validation of the steady-state response of the IPS model. Validating the dynamic response will probably prove much more difficult, as very little systemlevel simulation data is available in the open literature, beyond [17], for ships of this size and configuration. Simulation results indicate that the propulsion motors and associated converters will produce the largest thermal loads in the propulsion system. The analysis here demonstrates that secondary motor shutdown at cruise speeds can reduce waste heat given off by propulsion motors and converters by around 4%. A collateral benefit is reflected in significantly improved power factor of the primary propulsion motors. It is believed that reduced phase-set starting will also improve efficiency of the motor-converter pairs. [1] McCoy, T., Trends in Ship Electric Propulsion, 22 IEEE Power Engineering Society Summer Meeting, pp , 22. [2] Department of Defense (DOD), Militarily Critical Technologies List; Section 7: Energy Systems Technology, Office of the Under Secretary of Defense Acquisition, Technology, and Logistics, Pentagon, VA, September 25. [3] Holsonback, C and T. Kiehne, Estimation of System-Level Thermal Loads on the DDG1, Workshop on Transportable Megawatt Power Systems, March 27, Austin, Texas. [4] Rolls-Royce, Fact Sheet: MT3 marine gas turbine, online product documentation available at gasturbines/mt3/ downloads/mt3factsheet.pdf, 23. [5] TRAX Corporation, Analyst s Instruction Manual, Version 6.5.1, April 15, 24 and Programmer s Manual, dated March, 23. [6] Syntek Technologies, DD(X) Notional Baseline Modeling and Simulation Development Report in Support of the Electric Ship Research and Development Consortium, intra-consortium document, 1 August 23. [7] International Power Technology, Allison 51-KB7, online product documentation available at kb7_facts.htm, dated 27, accessed 12 April 27. [8] Davey, K., Ship Component In-Hull Optimization, Marine Technology Science Journal, vol. 39, n. 2, pp , 25. [9] Benatmane, M., T. McCoy, T. Dalton, and T. Cooper, Electric Power Generation and Propulsion Motor Development for US Navy Surface Ships, All-Electric Ship, Developing Benefits for Maritime Applications 1998, Session 5, Paper I, September 1998, pp [1] Lewis, C., The Advanced Induction Motor, 22 IEEE Power Engineering Society Summer Meeting, July 22, pp [11] Kotacka, R., Integrated Power Systems, 26 ONR Thermal Management Program Review, Berkeley, CA, oral presentation attended by Dr. Thomas Kiehne, 19 September 26. [12] Collins, M., Integrated Power Systems (IPS) for DDG 1 and Future Navy Ships, 26 Advanced Naval Propulsion Symposium, 26. [13] Norton, P. and J. Voyce, The Advancement of the Advanced Induction Motor (AIM) System, International Naval Engineering Conference (INEC) 22: The Marine Engineer in the Electronic Age, PowerPoint presentation, April 22. [14] Mercer, C., T. Dalton, E. Harvey, and M. Stauffer, Final Full Scale Advanced Development Test Results of the U.S. Navy Integrated Power System Program, Part D -- AES 23 Broadening the Horizons: New Ideas, New Applications, New Markets for Marine Electrical Technologies, February 23, pp [15] Wildi, T., Electrical Machines, Drives, and Power Systems, 5 th edition, Prentice Hall, Upper Saddle River, New York, 22. [16] Holsonback, C. Dynamic Thermal-Mechanical-Electrical Modeling of the Integrated Power System of a Notional All-Electric Naval Surface Ship, Master s Thesis, The University of Texas at Austin, May 27. [17] Andrus, M., S. Woodruff, M. Steurer, and W. Ren, Crashback Simulations of a Notional Destroyer-Class All-Electric Ship, 26 Advanced Naval Propulsion Symposium, 26.