INTEGRATED SIMULATION FRAMEWORK FOR CRASH BACK OPERATION

S. Brizzolara, P. Prempraneerach, G. Karniadakis and C. Chryssostomidis

MITSG 12-07

Sea Grant College Program Massachusetts Institute of Technology Cambridge, Massachusetts 02139

Project No. 2008-ESRDC-01-LEV

Integrated Simulation Framework For Crash Back Operation

S. Brizzolara P. Prempraneerach G. Karniadakis C. Chryssostomidis University of Genova Dept. of Mechanical Brown University, Sea Grant Dept. of Naval Architecture, Marine Engineering Division of Applied Mathematics, College Program and Electrical Engineering Rajamangala University of Providence, RI, USA, 02192 Massachusetts Institute of Technology Genova, Italy, 16145 Technology Thunyaburi, and also 292 Main Street, 02139 currently visiting MIT MechE Pathumthani, Thailand 12110 MIT Sea Grant College Design Lab Cambridge (MA) – USA [email protected] [email protected] [email protected] [email protected]

Keywords: All electric ship (AES), integrated power system must yield the most efficient power usage to maintain (IPS), hydrodynamics, multi-physics simulation, propeller continuity of service in the case of large power demand or modeling. critical situations, such as crash-back operations. A crash back is a standard emergency maneuver the board has to perform in the need of rapid stopping and Abstract reversing of the ship advance speed from the full head to the We formulate a modeling and simulation framework full astern condition. The rapid deceleration, induced by the that integrates models of an all-electric ship (AES) with ship propeller stopping and reverse motion, is one of the most and propeller hydrodynamic models. In particular we present violent maneuvers that can be imagined from different the first simulation study about the transient behavior of the perspectives: propeller and shaft dynamics and strength, propulsion system of AES during crash stop and backing transient loads on prime movers and powering supply maneuvers. A time domain model of all electric propulsion modules and even structural strength and vibrations in the system of a notional destroyer has been coupled with a stern structures above the propellers and thrust bearings. simplified transient model of the hull and propeller The complicated transient phenomena involved in the hydrodynamics. The integrated power system of the maneuver are regulated by the propeller and hull integrated power system (IPS) includes all main elements of hydrodynamics from one side and by the electric propulsion the chain, i.e. from power generators to control units down to system dynamics from the other. These dynamics have frequency controlled electric motors. The simplified unsteady completely different time scales, especially in the case of all hydrodynamic model of the hull and twin shaft lines with electric propulsion systems, but they can interact together fixed pitch propellers is based on semi-empirical quasi static conditioning the final result of the maneuver that can data corrected with complimentary CFD simulations. The contain harmful conditions for both systems. In particular transient behavior of the electrical power distribution and the logic of the torque and powering control of the electric motor control are analyzed and the relevant implications that drive is influencing the propeller hydrodynamics and vice these kind of violent transient maneuvers have on the versa. Some components as braking resistors are designed engineering of the main electrical components are outlined in just from loads coming out for these kind of maneuvers, so the paper. an effective design procedure must include correct

modeling of propellers and ship hydrodynamics. Recently 1. INTRODUCTION different CFD studies just on the unsteady propellers/hull In the new All-Electric Ship (AES) Integrated Power hydrodynamics during crash back maneuvers have System (IPS), there is an increasing demand for ship system appeared, as for instance in [Hur et al. 2011]. The authors automation, electrical weaponry, electric propulsion, and ship after concentrating on the modeling issues of the IPS are service distribution. The AES power system can be broadly here considering an integrated simulation framework that separated in five main sections: power generation, power can accurately solve both the IPS and ship/propeller conversion and distribution, propulsion loads, ship service hydrodynamic transients together. This work is the first step loads, and advanced system loads. Figure 1 gives a high-level towards this goal, featuring a simplified hydrodynamic description of the IPS. In particular, here we focus on the module, yet consistent in its main components, which can medium voltage DC system (MVDC) which is directly already give interesting indications to the designer, as it will involved in power generation and supply to the main be illustrated in the following sections. propulsion system. About 70% to 90% of power from the generator units in the fully integrated power system is consumed in the propulsion systems. The power distribution 2. ALL-ELECTRIC SHIP SIMULATOR The power generation unit using a 3-phase, 60-Hz, 21 MW The governing equations for all major components of a synchronous machine, driven by gas turbine, supplies MVDC system are described in [Marden et al. 2010], where electrical power to the induction motor, which acts as a first results were presented with very simplified propulsor for a full-scale USS DDG-51 Arleigh Burke- hydrodynamic models. The synchronous machine (SM) Class destroyer. The MVDC bus voltage is maintained at 5 produces 21 MW while the unidirectional rectifier converts kV by the type-AC1A exciter [IEEE 2006]. A detailed AC to DC voltage such that the nominal value of the MVDC description and model of individual components are given bus voltage is 5 kV. In particular, in the first generation of an in [Marden et al. 2010]. The command-following end-to-end simulator for AES we used a constant-slip comparison between constant-slip current control and direct approach to control the induction machine (IM), see Figure 1. torque control is performed using a ramp torque command For crashback operations the use of a proper AC drive from 0.110 to 0.359 MN-m and to 1.129 MN-m, applied to the induction machine is particularly important. corresponding to steady-state ship speed of 8, 15 and 20 The constant-slip current control optimizes the machine knots, respectively. To reach 15 and 20 knots ship speed, torque for a specified stator current by maintaining constant 3.502 and 8.199 MW total propulsion power is required, slip frequency. It has been used widely due to its respectively. Therefore, each of the two propulsion set implementation simplicity and robustness to disturbances. driven by the induction machine must correspondingly Nevertheless, magnetic saturation of the machine might occur produce about the half if this to propel the ship at these two in this control technique, thus the rotor flux likage must be forward speeds. The simulation results suggest that the kept below a threshold limit. Here we will use the direct direct torque control produces more torque ripple than the torque control approach for IM, which has also been widely constant-slip control since the inverter voltage produced by employed to control the induction machine because of its fast the switching table generates a lower harmonic frequency torque response to loads and supplied voltage changes, and compared to that produced by the current hysteresis. The its capability to operate with a four-quadrant AC drive. By propeller thrust has similar characteristics as the induction maintaining the magnitude and phase of the stator current machine torque. The ship speed is initially set at 8 knots and constant and choosing a proper inverter state from a ramps up to 15 and 20 knots within 150 and 275 seconds switching table, this technique can control the stator flux and respectively, as shown in Figure 4. torque directly and independently. In the classical direct torque control, two hysteresis controllers for controlling flux and torque are used to maintain the stator flux and the electromagnetic torque within both hysteresis bands. However, the hysteresis controllers are similar to a bang-bang control, containing discontinuous action, and as a result, the direct torque control produces large torque ripples.

Figure 2: Constant-slip current control : Induction Machine (IM) torque (first row), speed (second row), and stator flux (third row).

Figure 1: One line diagram of Medium Voltage DC Using a 3-phase, 15-Hz, 19 MW induction machine, the constant slip current control is compared to the direct torque control based on control characteristics and dynamics response, including a torque-command tracking and torque and current ripple. The results are shown in figures 2 and 3.

The induction motor shaft is directly coupled with a five- Figure 3: Direct-Torque Control : Induction Machine (IM) torque blade, fixed-pitch, highly skewed propeller [Boswell 1971]. (first row), speed (second row), and stator flux (third row).

capabilities has been tested. A recent study [Gheriani 2012] on an improved CFD model, based on a state of the art volume of fluid RANSE solver, has demonstrated that it is possible to obtain predictions of the total hull resistance with +/-3% uncertainty with respect to model test results on a wide speed range. Similar CFD models are actually being used in the continuation of the present study for the prediction of total resistance with appendages in forward and reverse speeds.

Table 1: Propellers used for maneuvering tests in model scale on DTMB5415 hull, given in [2] Figure 4: Propeller thrust (above) and ship speed (below) using both the constant-slip current control and the direct torque control.

3. HYDRODYNAMICS OF CRASH BACK

The dynamic model of the all electric propulsion system has been coupled with a preliminary transient model for ship and propeller hydrodynamics to test the functional characteristics of the main electrical components during crash Hull appendages actually vary between different back maneuvers. To the scope, the unclassified DTMB 5415 design studies made on this hull: in our case they consist of combatant hull was taken as reference in the preliminary tests a bow sonar dome, two pairs of bilge keels, a small skeg, of the simulator, as the closest public version of the notional twin shafts, struts, fairwaters and twin rudders with rudder- destroyer under consideration. In fact the hull, presented in shoes. Figure 5, is similar to the one actually built in the DDG-51 The resistance of the appendages was considered by series, being used for preliminary series of studies and the addition of a friction and form resistance components recently openly released in occasion of several workshops calculated with the exact Reynolds number and form factor about marine CFD [Various, 2010] and maneuvering of each appendage. For crash back simulations, the ship simulation [Various, 2008]. The full scale ship is a 142m x resistance needs to be estimated also in reverse speeds. 19m x 6m (LPPxBxT) guided missile destroyer and her full Preliminary CFD simulations of the hull, free to sink and load displacement measures about 8500tons. The hull has a trim, advancing in steady astern motion indicated that the pronounced sonar dome integrated in the bulbous bow and a bare hull resistance in these condition is nearly two times narrow stern transom, with a quite fine entrance body (low higher than the corresponding value at forward speed, when entrance angles) and a pronounced flare of the sections above including also the suction effect of the propellers. The final its design waterline. resistance curve assumed in the studies is given in Figure 6.

Figure 5: DTMB 5415 combatant hull whose hydrodynamic characteristics were taken in the simulations.

The prediction of the bare hull resistance in calm water has been subject of several CFD workshops [1] in which the Figure 6: Total Resistance curve estimated for the DTMB5415 accuracy of different types of RANSE (Reynolds Averaged hull with appendages in forward and steady forward and reverse Navire Stokes Equations) solvers with two phase flows motion

Main geometrical data of the different propellers used for Consistently also the non-dimensional thrust and maneuvering tests in model scale are given Table 1. A fixed torque coefficients are referred to the undisturbed speed pitch propeller with the same pitch ratio of that selected by relative to the 0.7R blade section, in such a way: FORCE but with increased expanded area ratio of AE/AO=0.7 and a smaller hub diameter d /D=0.18, was assumed in order H ( ( ) ) ⁄ { ( ) } to be more similar to values usually taken by modern propeller designs of high speed destroyers.

( ( ) ) ⁄ { ( ) }

in which T is the thrust, Q the torque, n the rpm and D the diameter of the propeller; VA is the ship advance speed defined in terms of the wake fractions as VA=V·(1-w).

Table 2: Quadrant definition for  CT* CQ* coordinate system

Figure 7: Predicted Propeller thrust and torque curves in the first quadrant with the design operating point The final propeller CT, CQ,  diagram used in this A check of the propeller efficiency and design point study is presented in Figure 8, as estimated on the basis of a (indicated with a yellow bullet) made with Wageningen B4.70 propeller, tuned in such as way to match the thrust standard family of propellers [Kuiper 1992] gave the results needed by the hull at the design speed of 30 knots. presented in Figure 7: at the design speed of 30 knots, the It is worth noting that some propeller geometric propeller has a rotational speed of about 122 rpm and a quite characteristics have a relevant influence on the maximum good open water efficiency (about 0.72), in line with most propeller thrust and torque during crash back maneuvers. recent navy ships of this size. As usual in the diagram of The analysis of the B-series propeller tests results in the Figure 7 propeller hydrodynamic characteristics are given in four quadrants [van Lammeren et al. 1969], it can be argued 2 4 that the expanded area ratio has a major influence on the terms of non dimensional thrust coefficient KT=T/(0.5n D ) * * and torque coefficient K =Q/(0.5n2D5) as a function of the maximum CT and CQ that a fixed pitch propeller generate Q in the diagram, while the pitch ratio has a prevalent effect advance ratio J=VA/nD, which is regulates the actual load on * the propeller blades. on the maximum CQ only in stopping and backing. The ordinary differential equation which describes the Indeed, also the actual blade design in terms of chord, dynamic of the ship advancing in rectilinear decelerated skew and rake distribution along the radius is to be motion needs the information about the time depending considered for the correct estimation of the propeller forces propellers thrust. For stopping and backing operation the during these transient maneuvers. developed thrust and requested torque by the propeller are For the above reasons, it is desirable to use the exact conveniently represented, in steady condition, by the so designed propeller for the calculation of the four quadrant called four quadrant propeller diagrams, which rely on a diagrams needed in the final assessment of propulsion different definition of the advance coefficient. The new transient behavior. In this respect Numerical flow solvers coefficient considers the different combinations of positive [Luo et al. 2010] [Brizzolara et al. 2008] able to fully and negative ship and propeller speeds in terms of the consider separation and turbulence, seems to be a valid tool to obtain accurate predictions. A validation study is in fact advance angle  calculated for a cylindrical blade section currently under completion. made at 0.7R (propeller radius). In formulae: So simulator integrates the ordinary differential equation for the speed of the ship advancing in rectilinear ( ) ( ) decelerated motion, and it is considering in it the variable The correspondence of each quadrant with the different propeller thrust deriving from the ship speed and propeller ship accelerating and stopping maneuvers is given in Table 2. rpm changes during the maneuver. An example of the values taken by the advance angle  during a crash stop

maneuver with constant torque set to one third of the synergistically dissipate the surplus energy, as shown in maximum torque in forward speed, is presented in Figure 9. Figure 10. Different resistance values have been investigated to minimize the instantaneous power dissipation using the braking chopper.

Table 3: Oil cooled power resistor from GINO AG (http://www.gino.de/download/download/71_prospekt_oil_cooled_power_resistors_e ngl.pdf) Continuous Current Voltage up Oil Volume (l) Load (kW) (A) to (kV) 50 400 10 145 100 400 12 205 150 630 12 355 270 630 12 500 400 630 24 775 540 630 24 1080 800 630 24 1600 Figure 8: Torque and Thrust Coefficient curves estimated for the assumed propeller A crash astern scenario is performed to examine the interaction among all different electric machines and Due to the completely different time scales between the components as well as to evaluate the power consumption two dynamics, propeller and ship, the quasi steady approach and regeneration. Moreover, different values of three- followed to estimate propeller thrust and torque seems well parallel braking resistors are compared to minimize power justified. An estimate of the ship added mass for the surge dissipation in each individual resistor. For the crash astern motion is instead considered, using classical relations. operation, as previously described, the propeller hydrodynamics is directly coupled and driven by the induction machine. The results of these tests are shown in figures 11-16.

Figure 9: Advance angle change during the crash stop maneuver with given constant toque on the propellers.

4. CRASH ASTERN OPERATION To avoid an accumulation of heat in the braking resistors during the crash astern operation, a braking chopper Figure 10: Three parallel braking resistors are controlled by the is added to maintain the MVDC bus voltage at maximum braking chopper to maintain the MVDC bus below the maximum threshold level of 6 kV. If the MVDC bus voltage rises above threshold voltage of 6 kV. The braking chopper is operated at 100 this maximum threshold, the braking chopper turns on the Hz. crowbar to draw current through the braking resistors with a Initially, the ship speed is kept at 15 knots and then switching frequency of 100 Hz. According to Table 3, there increased to 25 knots within 50 seconds. Subsequently, the exist braking resistors in the market with oil cooling system induction machine undergoes the braking operation by of various power ratings and maximum voltage, which can be applying a ramp negative torque to the propeller shaft from applicable to the MVDC integrated power system. In this 981.1 kN-m to -0.4 MN-m within 25 seconds. The case, three braking resistors are connected in parallel to rotational speed of the propeller or motor shaft gradually

decreases and reaches zero speed in about 43 seconds, as shown in figure 12. During this braking period, the ship still cruises in forward direction [Hur et al. 2011], see figure 14, and the induction motor operates as a generator that supplies the regenerated power back to the MVDC bus, which causes the gas turbine (GT) and synchronous machine (SM) to become idle or SM torque drops to zero in figure 11. In addition, the MVDC bus voltage rises and limits at the maximum threshold of 6 kV by switching on-off the braking chopper, shown in figure 13, thus a large instantaneous current is drawn from the MVDC bus current intermittently. The synchronous machine phase current decreases to zero Figure 13: Bus voltage and current (first and second rows), SM during this period. This regenerative power must be phase current (third row), and IM phase current (fourth row). consumed by the three-parallel braking resistors of 75 Ω for a short period of 24.6 second. Once the rotational speed of the induction machine (IM) becomes negative, the IM again operates in the motor mode, drawing power from the MVDC bus, at 93.4 seconds. As a result, the MVDC bus voltage suddenly drops and the gas turbine and synchronous machine start to supply more energy to the MVDC bus once again after idling.

Figure 14: Propeller thrust (first row) and ship speed (second row).

To minimize the dissipated power in each braking resistor, five different resistance values of 15, 30, 45 60 and 75 are compared in terms of peak instantaneous power and average power dissipations, as shown in Table 4. Figures 15 and 16 show the time histories of the peak electrical power drawn from the MVDC bus for two representative cases during the Figure 11: Gas Turbine (GT) torque (first row), Synchronous Machine (SM) torque and rotational speed (second and third rows). regenerative duration. As a result, the larger value of braking resistor, the smaller the peak instantaneous power dissipation. Therefore, the existing power resistor from Table 3 could be employed when the braking resistor size is above 45 Ω. However, the average power dissipation over the regenerative period for all resistor values is comparable at 4.8-4.9 MW.

Table 4: Average and peak instantaneous power dissipation for different resistance values of three-parallel braking resistors Individual 15 30 45 60 75 Resistor (Ω) Average Power (MW) 4.90 4.78 4.80 4.81 4.81 Figure 12: Induction Machine (IM) torque (first row), speed Peak Instantaneous Power 2.41 1.22 0.82 0.62 0.50 (second row), and stator flux (third row). (MW)

REFERENCES Borda G. (1994). Resistance Powering and Optimum Rudder Angle Experiments on a 465.9ft Guided Missile Destroyer (DDG-51) Represented by model 5415-1 and Fixed Pitch Propellers 4876 and 4877. Boswell R.J. (1971) Design, Cavitation Performance, and Open- Water Performance of a Series of Research Skewed Propellers”, Report 3339, Department of the NAVY, Naval Ship Research and Development Center. Boulghasoul Z., Elbacha A., Elwarraki E., Yousfi D., (2011) Combined Vector Control and Direct Torque Control and experimental review and evaluation”, International Conference on Multimedia Computing and Systems, 2011

Figure 15: Electrical and mechanical instantaneous powers for (ICMCS), pp. 1-6. individual braking resistor of 15 Ω or equivalent resistance of 5 Ω Brizzolara S., Villa D., Gaggero S. (2008). A systematic from three-parallel resistors. Comparison between RANSE and Panel Method for Propeller Analysis. Hydrodynamics VIII, 8th International Conference on Hydrodynamics, ICHD 2008. Nantes, Oct. 2008, vol. 1, p. 289-302 Gheriani E. (2012) Fuel Consumption Prediction methodology for early Stages of Naval Ship Design. MSC thesis in Naval Architecture and Marine Engineering, MIT MechE, C. Chryssostomidis supervisor. Hoang Le-Huy (1999) Comparison of field-oriented control and direct torque control for induction motor drives, Thirty- Fourth IAS Annual Meeting. 1999 IEEE, Industry Applications Conference, Vol. 2, pp. 1245-1252. Hur J.W., Lee H., Chang B.J. (2011). Propeller Loads of Large Commercial Vessels at Crash Stop. Second International Symposium on Marine Propulsors smp’11, Hamburg, Figure 16: Electrical and mechanical instantaneous powers for Germany, June 2011. individual braking resistor of 75 Ω or equivalent resistance of 25 Ω IEEE Std 421.5-2055 (2006) IEEE Recommended Practice for from three-parallel resistors. Excitation System Models for Power System Stability Studies, pp. 1-85. (Revision of IEEE Std 421.5-1992) 5. CONCLUSIONS Krause P.C., Wasynczuk O., Sudhoff S.D. (2002) Analysis of We have presented for first time the performance of an end- Electric Machinery and Drive Systems, Second Edition, to-end simulator of the all-electric ship for the MVDC system IEEE Press and Wiley-Interscience, 2002. in the cases of crash astern ship maneuvers, based on the Kuiper, G, “The Wageningen Propeller Series”, 1992, MARIN integration of a simplified propeller/hull hydrodynamic Publication No. 92-001 model. In particular, we designed two different control Lammeren, W.P.A. van, Manen, J.D. van, Oosterveld, M.W.C., strategies for the induction motor and we also investigated “The Wageningen B-Screw Series”, Transactions SNAME, Vol. 77, pp. 269-317, 1969. the possibility of using different resistor elements for a Luo X., Chryssostomidis C., Karniadakis G.E. (2010) Spectral smooth operation of the MVDC system during the crash element/Smoothed profile method for turbulent flow astern operation. Future work will focus on a more accurate simulations of waterjet propulsion systems, Proceedings of representation of the propeller/hull hydrodynamic during Grand Challenges in Modeling & Simulation, Simulation crash astern maneuvers, by means of complementary Multi-Conference, July 2010, Ottawa, Canada. simulations with state of the art viscous solvers; and on the Marden M. M., Prempraneerach P., Karniadakis G.E., stress calculation on the propeller blades during the crash Chryssostomidis C. (2010) “An End-to-End Simulator for astern operation so that we do not exceed the maximum yield the All-Electric Ship MVDC Integrated Power System”, stress of the propeller blade. Future studies will determine the Summer Simulation Multi-conference (SummerSim). Menard L.P.M. (2010) Prediction Performance and Maneuvering maximum slope that we can impose on the IM torque Dynamics for Marine Vehicles Applied to DDG1000. MSc command. Thesis in Naval Architecture and Marine Engineering, MIT MechE, M. Triantafyllou Supervisor. 6. ACKNOWLEDGMENTS Various (2010). http://www.gothenburg2010.org/, Marine CFD This work was supported by the Office of Naval Research Workshop 2010, Gothenburg, Sweeden. Various (2010). http://www.simman2008.dk/5415/combatant.html (N00014-08-1-0080 ESRD Consortium). SimMan 2008 Maneuvering Workshop, Lynby, Denmark