Propeller - diesel engine interaction in a turn

Transcription

1 Propeller - diesel engine interaction in a turn Lt P J M Schulten, MSc, MiMarEST, Ceng Royal Netherlands Naval College S L Toxopeus, MSc Maritime Research Institute Netherlands D Stapersma, MSc, FIMarEST, CEng Royal Netherlands Naval College, Delft University of Technology SYNOPSIS In this paper the simulation results of a ship mobility model are presented. This model consists of an elaborate propulsion model (diesel engine, shaft, propellers), a modified propeller model and a ship manoeuvring model (hydrodynamic forces, ship standard manoeuvres). The objectives are to investigate the interaction between the diesel engines and the propellers in a turn and to explore the value of a mean value diesel engine model. The results are compared with the results of the validated manoeuvring model FreSim. This model does not have a sophisticated diesel engine sub-model but a simple constant torque propulsion engine. From the simulation results, three conclusions can be drawn. Firstly, the elaborate diesel engine model calculates the engine torque in such a way that the manoeuvring results (ship speed, rate of advance etc) are comparable with the validated FreSim results. Secondly, apart from engine torque, the diesel engine model also calculates other values such as exhaust gas temperature. This exhaust gas temperature can reach unexpected high values in a turn. Finally, in a twin-shaft configuration, the inner and outer propellers are loaded differently because of differences in wake. This results in difference in loading and exhaust gas temperature of the diesel engines. INTRODUCTION In marine engineering, a broadening cooperation between industry, research institutes and academia can be observed. This is the result of a more general trend: an increasing interest in the total behaviour of complex systems and of the behaviour of individual components within this total system. One of the ways to investigate such total system behaviour is by creating an overall model in which the various scientific areas bring in their proven knowledge with more or less equal importance. The case study of this paper is a total model of the ship mobility system. It consists of a complex diesel engine model, a state of the art ship manoeuvring model and a propeller model in which the effects of oblique flow are Author s Biography Paul Schulten graduated from the Royal Netherlands Naval College in 1997 and obtained a Master of Science degree in Mechanical Engineering form the Delft University of Technology in From 1997 to summer 2000 he was stationed at HNLMS De Ruyter as Deputy Maintenance Engineering Officer and sailed both to the Caribbean and to the Orient. Since September 2000 he is stationed at the Royal Netherlands Naval College to lecture Fluid Mechanics and to perform a PhDresearch on the dynamics of the ship movement system. Paul Schulten is a member of the Institute of Marine Engineering, Science and Technology. Serge Toxopeus graduated in 1996 at Delft University of Technology, Faculty of Mechanical Engineering and Marine Technology with a Master of Science degree in Naval Architecture. Since that time, he has been employed at MARIN as a project manager in the field of ship hydrodynamics, specializing in ship manoeuvring. Besides managing model test projects and the development of generic mathematical manoeuvring models and manoeuvring simulation programs, his current activities are directed at the development of practical application methods of viscous flow calculations for manoeuvring ships. Douwe Stapersma, graduated in 1973 at Delft University of Technology in the field of gas turbines and then joined NEVESBU - a design bureau for naval ships - and was involved in the design and engineering of the machinery installation of the Standard frigate. After that he co-ordinated the integration of the automatic propulsion control system for a class of export corvettes. From 1980 onward he was responsible for the design and engineering of the machinery installation of the Walrus class submarines and in particular the machinery automation. After that he was in charge of the design of the Moray class submarines in a joint project organisation with RDM. Nowadays the author is professor of Marine Engineering at the Royal Netherlands Naval College and of Marine Diesel Engines at Delft University of Technology. Douwe Stapersma is a fellow of the Institute of Marine Engineering, Science and Technology.

2 included. Using this model, not only the manoeuvring properties of the ship, but also the behaviour of the diesel engines during a turn can be investigated thus linking the world of the hydrodynamical engineer and the mechanical / marine engineer. In the first part of the paper the model concept is presented and the various sub-models are briefly explained. Furthermore, reference is made to literature where specific details of the sub-models can be found. In the second part of the paper the results of two simulations are analysed. Finally, the objective of this paper is twofold. Firstly, a case study like this gives insight in the possibilities and limitations of the total-modelling concept. Secondly, if the link between the hydrodynamical and mechanical world is better understood, the operational possibilities and limits of the ship mobility system can be better explored and optimised by better control. THE SHIP MOBILITY SYSTEM In figure 1 the concept of the ship mobility model is presented. The model consists of five sub-models: Ship movement. In this sub-model the equations of motion are modelled. Based on the forces (inputs), the model calculates the accelerations, velocities and positions of the ship. Active forces. This sub-model calculates the forces that actively move the ship. The ultimate outputs of the model are thrust forces generated by the propellers, thrusters, waterjets etc. Also the subsystems that are needed to drive the propulsors are included. These subsystems are the main engine, the gearbox etc. Passive forces. The passive forces are the forces that exist because of a relative velocity difference between the water and the hull with its actuators such as rudder(s) and stabilisers. Based on the velocities and rudder position (the inputs), the hydrodynamic coefficients and forces are calculated. Disturbances. In this block the effects of wind and waves (the uncontrollable inputs) are modelled. Mobility control system. This system controls the movement and propulsion of the ship at the highest level. Various controllers also are present at other levels of the ship mobility system [1]. multiple input MOBILITY CONTROL SYSTEM ACTIVE FORCES propulsion rotor dynamics propulsor SHIP MOVEMENT surge yaw sway roll pitch heave PASSIVE FORCES rudder forces hull forces DISTURBANCES wind waves Fig 1 Ship mobility model concept. In the case-study presented in this paper the five sub-models have been implemented in Matlab-Simulink in such a way that a twin-shaft diesel engine-propeller driven ship in a turn is simulated. In the following sections, some specifics of diesel engine model and the propeller model (active forces), and the hull and rudder forces (passive forces) are given. The basic equations used in the ship movement sub-model can be found in [2]. Finally, the disturbances and the mobility control system are not regarded in this study. The inputs of the model are desired shaft speed and rudder angle, which are directly fed into the active and passive forces sub-models. An investigation in the possibilities and limitations of various mobility control regimes is presented in [3].

3 Diesel engine model The diesel engine model used is a mean value diesel engine model. Figure 2 shows the concept of such a model. The diesel engine is modelled as a network of volumes and resistances. In a resistance (e.g. the airfilter in figure 2) the mass flow and flow temperature are calculated as a function of the pressure difference over the element using the momentum and energy equations. In a volume (e.g. the inlet receiver), the pressure and temperature are calculated as a function of net mass flow in the volume using the mass balance and the first law of thermodynamics. The modelling concept as shown in figure 2 applies to both the filling and emptying and the mean value diesel engine models. The difference between both models is the way the cylinder process is treated. In the filling and emptying models, the thermodynamical processes in the cylinder are regarded at each crank angle. To be able to do this a sophisticated combustion model is needed. The flow through the cylinder is calculated in the same way as in the other elements of the network. fuel n eng CYLINDER M eng af Inlet volume com Air cover cac Inlet receiver cyl in Cylinder volume cyl out Outlet receiver tur Silencer volume sil T af T iv T com T ac T T cac T civ T ir ir T cov T T or tur T sv p cyl T amb p amb Airfilter p iv Com p ac Charge Air Cooler p ir Cylinder Inlet valves Cylinder Outlet valves p or Turbine p sv Silencer T sil p amb n turbo M com + - M tur n turbo 1 2 I Fig 2 Diesel engine model concept In a mean value model the cylinder process is divided in a number of discrete processes: inflow, combustion, scavenging, blowdown, outflow etc. For each of these processes mean values of mass flow, temperature, composition, pressure and work are calculated using only algebraic functions. The various mass flows are then thermodynamically mixed in the outlet receiver resulting in the entry condition for the turbine. Two remarks regarding the use of pure mean value versus filling and emptying models have to be made. Firstly, the simulation time for a filling and emptying model by definition is longer since the time step of such a simulation is in the order of one degree crank angle. The time scale of mean value models is in the order of one engine revolution, an advantage especially if the diesel engine model is part of a larger model as is the case in this paper. Secondly, in case of the mean value model the algebraic relations in fact are pre-fabricated solutions of the differential equations describing the processes in the cylinder. Although a mean value diesel engine model does not provide information on a crank angle scale, it still predicts the same information in terms of number of signals. Examples are: mass flow, turbocharger speed, inlet receiver pressure and temperature, maximum cylinder pressure and temperature, outlet receiver pressure and temperature, engine torque, exhaust gas temperature. Further details of the diesel engine model can be found in [4] and [5]. Passive forces: FreSim The passive forces are calculated by FreSim, a simulation program developed by the Maritime Research Institute Netherlands (MARIN). This program was developed to simulate the manoeuvrability of high-speed surface ships. The theory is based on the cross-flow drag theory as described by Hooft et al [6], [7]. The simulation model is capable of predicting the motions of the vessel in calm water conditions in 4 degrees-offreedom, i.e. surge, sway, roll and yaw. The forces on the ship s hull are calculated using a full non-linear model. The linear parts of the forces are determined by empiric formulae based on the main particulars of the ship, while the non-linear part also includes the influence of local hull details such as the local sectional draught. Currently, studies are conducted at MARIN regarding the use of the so-called slender body method in order to include local hull details into the linear coefficients as well, see e.g. Keuning et al [8]. This means that already the impact of minor changes in the hull design can be verified using the simulations and that the method will be applicable to a broader range of generic

4 ship types. Further studies are also focussing at using viscous flow calculations to generate the hydrodynamic coefficients that can be used to predict the forces on the ship. Ultimately, this will provide the most generic approach in calculating the forces on the ship based on the complete hull form geometry. A rich data-set of full scale and model scale results of standard manoeuvres is available at MARIN which can be used for validation exercises and full scale - model scale correlation studies. New automated benchmarking tools have been developed to extensively check the validity of manoeuvring prediction software. Currently the benchmark database for FreSim contains full-scale and model-scale manoeuvring results for 15 different relevant high-speed slender vessels. The manoeuvring results consist of parameters derived from well-known IMO type manoeuvres: turning circle, zig-zag and stopping manoeuvres. Figure 3 presents an example of turning circle benchmark results of the tactical diameter and the nondimensional rate of turn stc, the latter being defined by: γ stc rstc L pp (1) U stc with r stc and U stc the rate of turn and velocity in the steady turn respectively. In these graphs each point represents a comparison between an experiment (full scale indicated by crosses or model scale indicated by squares) and a prediction. The value on the horizontal axis represents the experimental data and the vertical axis represents the prediction. Ultimately, all data points should be located on the diagonal line for a perfect prediction. From this figure, it is concluded that a reasonable correlation is found between the simulation results and the measurements. More details of the validation of FreSim including the benchmark results of zig-zag manoeuvres are presented by Loeff and Toxopeus [9]. Fig 3 Tactical diameter and non-dimensional turning rate validation of FreSim Propeller model Instead of a straightforward application of Kt- and Kq diagrams, two aspects of a propeller in a turn are incorporated in the applied propeller model. Firstly, the transverse velocity of the ship results in oblique inflow in the propeller. This means the Kt- and Kq- diagrams cannot be directly used (they are based on purely axial inflow) but have to be adapted. Gutsche [10] presented an analytical treatment of the propeller with oblique inflow and this theory is the basis of the propeller model. The second aspect of a propeller in a turn is the fact that the wake in a turn is different from the wake in a purely longitudinal ship motion. Based on full-scale measurements on a slender naval vessel Kuiper [11] found that the axial inflow is unaffected in a turn while the drift angle generates a transverse velocity component in the lower half of the propeller plane. Lower half inner propeller Lower half outer propeller V a nd nd V a V tr V tr Fig 4 Resulting inflow velocities and angle in the lower half of the propeller plane.

5 In figure 4 the effects of the transverse velocity for a twin-propeller ship in a port side turn are shown. Because the inner and outer propeller turn in a different direction (in this case inward over the top), the effects of this transverse velocity in the lower halfs of the propeller planes are different: for the outer propeller, the net inflow velocity and entrance angle of the blades is enlarged. For the inner propeller they are decreased. This results in a difference in torque and thrust between the inner and outer propeller. In the upper half of the propeller plane the effects of the transverse velocity are the opposite but Kuiper found that the transverse velocities are much smaller in this upper half resulting in a smaller torque and thrust difference between the inner and outer propeller. In the end, a net torque and thrust difference between outer and inner propeller remains. Both aspects, oblique inflow and varying wake (especially the transverse velocities), also result in the fact that the net inflow speed and angle vary during one revolution for both the inner and outer propeller. This is caused by the fact that during one revolution the direction of the axial and transverse velocities remain constant, while the direction of the circumferential speed ( nd in figure 4) rotates. This means that the advance coefficient J, the input to the Kt and Kq diagrams, also varies. The Gutsche-theory makes it possible to use the regular Kt and Kq diagrams by calculating the oblique advance coefficient as a function of propeller angle. The final results are torque and thrust that vary during one propeller revolution from which mean torque and thrust can be calculated. This is done by performing an analytical / algebraic calculation at each time step. In fact, this is comparable with the mean value strategy that is used in the diesel engine model. SIMULATIONS Two simulations were carried out using the total model as described above. The inputs of the total model are the shaft speed setpoint and the rudder angle setpoint. In the diesel engine governor the shaft speed setpoint is translated to fuelrack position by comparing the setpoint with the true value using a PI-controller. In the simulations the actual rudder angle is the same as the rudder angle setpoint and is kept constant. This means the rudder hydraulics do not have to be included. In the first simulation the shaft speed setpoint is 100% rpm, being the maximum setpoint when sailing on cruising diesel engines. The rudder angle setpoint is 35 deg (maximum rudder). In the second simulation the shaft speed setpoint is 74% of the maximum rpm and the rudder angle setpoint again is 35 deg. Apart from the simulations with the total model also simulations with the original FreSim program were executed. In this program the engines are considered to be constant torque engines. Furthermore, the propeller and the effects of oblique inflow are incorporated in a different way. By comparing the results of the total model with the original FreSim model the advantages of a sophisticated diesel engine model over a constant torque assumption can be investigated. Furthermore, since the original FreSim program is well validated, comparison gives insight in the validity of the total model. Ship manoeuvring Figure 5 shows the normalised longitudinal velocity u/u eq, the normalised ship transverse velocity v/v eq, the normalised drift angle eq and the normalised rate of turn r/r eq. The values are normalised using the equilibrium values at the end of the turn of the 100 % rpm FreSim simulation. In each graph four lines are displayed, the thin lines represent the FreSim results at 100% rpm and 74% rpm, the thick lines are the results of the total model at 100% rpm and 74% rpm, It can be seen that the loss of longitudinal velocity is greater in the FreSim model. This is due to the fact that in the total model a different propeller is used. The reason for this is that in the FreSim program only a limited number of standard propellers is available while in the total model the exact Kt and Kq diagrams of the propellers are used including the effect of oblique motion on the advance ratio. Since both propellers have the same torque as will be shown in figure 6, the thrust can and will be different, resulting in a different longitudinal velocity. The differences in the transverse velocity are much smaller while rate of turn and the drift angle are nearly the same in both models.

6 Fig 5 Normalised velocities and driftangle for FreSim and the total 100% and 74% rpm. In figure 6 the normalised shaft speed n/n 0, the normalised propeller thrust T/T eq, the normalised torque Q/Q eq and the normalised ship trajectory x 0 /L, y 0 /L are presented. The propeller thrust and torque again are normalised using the equilibrium values of the 100% rpm FreSim simulation. The shaft speed is normalised using the initial shaft speed of the 100% rpm FreSim simulations. Finally, the trajectory is normalised using the ship length. From these graphs an important difference between the 100% rpm and 74% rpm simulations can be observed: in the 100% rpm simulation the shaft speed decreases while in the 74% rpm simulation the shaft speed remains constant. This is caused by the fact that in the 100% rpm simulation the diesel engine operates at its maximum fuel rack position as will be shown in figure 7. Initially, the propeller torque will rise because of the turn. The engine torque cannot rise because of the maximum fuel rack position and as a result the shaft speed will decrease. Because of this the engine torque and propeller torque will decrease as well until equilibrium is reached. In the 74% rpm simulation the maximum fuel rack position has not been reached and an increase in propeller torque will be followed by an increase in the engine torque. The engine is able to maintain the shaft speed. Fig 6 Normalised shaft speed, propeller power and torque and ship trajectory for FreSim and the total 100% and 74% rpm.

7 Diesel engine Figure 7 shows the normalised inputs of the diesel engine model: engine speed (normalised with initial 100% engine speed) and fuel rack position (normalised with the equilibrium fuel rack position). Also presented are the main output of the model (engine torque normalised with the 100% rpm equilibrium torque) and an example of an internal engine variable: the exhaust gas temperature, again normalised with the equilibrium value. In the 100% simulation the engine torque initially will be different for the inner and outer propeller: a higher torque for the outer propeller due to the differences in wake. Because of this the engine speed of the outer propeller will decrease more resulting in a lower equilibrium engine speed. As argued before, the engine speed in the 74% rpm simulation will remain constant. Since the torque of the outer propeller is higher, the engine torque will be higher as well at the cost of a higher fuelrack position. The exhaust gas temperature shows an interesting profile. In the 100% rpm simulation the exhaust gas temperature will remain practically constant while in the 74% rpm simulation the exhaust gas temperature reached a level that is considerably higher. This is caused by the fact that in part load the air excess ratio is lower. Fig 7 Normalised engine speed, fuel rack position, engine torque and exhaust gas temperature for the total 100% and 74% rpm. Summarising, because of the differences between the inner and outer propeller and because of the differences between the 100% rpm simulation (constant fuel rack) and the 74% rpm simulation (constant shaft speed), two observations can be made. 1. Both in the 100% and the 74% simulations, the outer propeller is loaded higher than the inner propeller. In the 100% simulation this results in a constant engine torque and a decreasing engine speed. In the 74% simulation this results in constant engine speed and increasing torque. 2. In the 100% rpm simulation the engine exhaust temperature remains nearly constant. In the 74% rpm simulation the exhaust temperature rises considerably to a level higher than the level in the 100% rpm simulation. Furthermore, there is a difference between the port and starboard engine. CONCLUSIONS The ship mobility model as described in this paper is built by the first author as part of a PhD-research. The objective of this research is to investigate the interaction between the diesel engines and the propellers during manoeuvring. In this paper some preliminary results have been presented. From these results it seems clear that the addition of a sophisticated diesel engine model and a sophisticated propeller model to a state-of-the-art ship manoeuvring model does not affect the overall ship manoeuvring velocities. The only noticeable difference arises in the ship longitudinal speed but this is caused by the different propellers that are used. Although figures 5 and 6 suggest that results of the total model are good in comparison with the validated FreSim model, real validation still has to be performed. This will be done in the final stage of the PhD-research. The added value of the diesel engine model is a vast amount of relevant data that can be produced by this model. The validity of this data is being investigated at this time, again as part of the PhD-research. As an example of relevant data, apart from the engine torque also the exhaust gas temperature for both the starboard and port

8 engine is shown, as part of a general research into thermal overload of diesel engines. From the engine torque and the exhaust gas temperature it is clear that the diesel engine behaviour is strongly influenced by two aspects: whether the maximum fuel rack position is reached or not and whether the engine is connected to the inner or the outer propeller in a turn. Because of this, especially the exhaust gas temperature shows an interesting behaviour: the temperature in a low speed turn is higher than the temperature in a high speed turn. REFERENCES 1. Paul Schulten, Douwe Stapersma and Arthur Vermeulen, The Classification of a Ship Mobility Model, Proceedings of the 13 th Ship Control System Symposium, 7-9 april 2003 Orlando, USA. 2. M A Abkowitz, Stability and Motion Control of Ocean Vehicles, Massachussetts Institute of Technology, SBN , Douwe Stapersma, Hugo Grimmelius and Paul Schulten, A Fresh View on Propulsion Control, Proceedings of the 2004 INEC congress Amsterdam, Institute of Marine Engineering, Science and Technology, Paul Schulten and Douwe Stapersma, Mean Value Modelling of the Gas Exchange of a 4-stroke Diesel Engine for Use in Powertrain Applications, Society of Automotive Engineers, SEA , presented at the SAE 2003 World Congress, Detroit, USA. 5. Hugo Grimmelius, Dave Boëtius and Patrick Baan, The influence of sequential turbocharging control on propulsion behaviour, Proceedings of the 12 th Ship Control System Symposium, The Hague, J.P. Hooft and F.H.H.A. Quadvlieg, 'Non-linear hydrodynamic forces derived from segmented model tests', MARSIM 1996, ISBN J.P. Hooft and J.B.M. Pieffers, 'Maneuverability of Frigates in Waves', Marine Technology, Vol. 25, No. 4, pp , October Keuning, J.A.; Toxopeus, S.L. and Pinkster, J. "The Effect of Bowshape on the Seakeeping Performance of a Fast Monohull". Proceedings of FAST 2001 Conference, Southampton, September G.B. Loeff and S.L. Toxopeus, 'Manoeuvring Assessment in Concept Ship Design', Proceedings of NAV 2003 symposium, Palermo, June F Gutsche, The study of ships propellers in oblique flow, Defence Research Information Centre, Schiffbauforschung, 3, 3/4 (1964) , G. Kuiper, M. Grimm, B. McNeice, D. Noble, M. Krikke, Propeller Inflow at Full Scale During a Manoeuvre, proceedings of the 24 th symposium On Naval Hydrodynamics, Kuruoka, Japan, 2002.