2. A variety of control schemes especially for dealing with situations of heavy weather and rough sea have been investigated. The following scenarios/situations have been examined in detail:
a) Overspeed during fast load sinks: For dealing with this major problem which occurs mainly when engine is ran near-MCR under heavy weather, the following two techniques have been used superimposed:
- Reduction of the I-term gain for positive acceleration
- Torque prediction feed forward implemented as a speed setpoint reduction when a major load sink is foreseen to occur.
A fast and large load sink can occur under extreme weather conditions in combination with other events such as partial propeller emergence.
b) Excessive speed during slow, quasi-periodic load fluctuation: A special adaptive speed setpoint technique has been used that reduces the maximum speed observed when the engine is running near-MCR under rough sea conditions. Excessive speed (i.e. long time intervals with rpm slightly above MCR) occurs due to the quasi-periodic, low-frequency fluctuation of propeller load demand.
c) Heavy running due to weak load variations: Engine heavy running occurs mainly when the ship is sailing in heavy weather and is characterized by rpm excursions far from the requested setpoint. This occurs in combination with increased activation of the control limiters, i.e. the non-linear part of engine control that disables the smoother running provided by the linear PI speed regulator. In order to reduce these negative effects, the technique of "soft limiters", i.e. limiters applied to the speed setpoint instead of the fuel index, in combination with raised index limiter curves, is used [4]. This technique guarantees less activity on behalf of the non-linear part of engine control, without any compromise on safe plant operation.
d) Potential turbocharger stall: The fact that the turbo charging system and engine overall inertias are different, as well as, the time lag due to the discrete nature of the engine-turbocharger interaction (through the exhaust valves and inlet ports), may lead to compressor surge [5][6][7][8]. Turbocharger stall leads to a major disturbance to air delivery to the cylinders and consequently in extreme cases can result in combustion and thermal loading problems in the engine [7]. In the ACME project, detailed investigation has shown that if a fuel gradient (time derivative) index limiter is used with the turbine outlet temperature parameterising the limiter curve, risk of stall is reduced during fast changes of engine fuelling. This kind of changes in engine fuelling may be due either to a large, fast propeller load sink or to demand of fast acceleration/deceleration of the plant.
The follow-up to the theoretical simulation effort was to build:
1. A Torque Prediction (TP) System, i.e. a shipboard, real-time computer system framed by the required sensors and ND interface modules implementing the Propeller Torque Demand Prediction Algorithm.
2. A digital Governor/Engine Control Unit (ECU) based on a Motorola 68332 CPU and implementing the above control schemes for engine running in heavy weather in combination with the standard PI speed regulating function realised by any conventional engine governor.
Both the ACME systems were integrated to the existing control installation onboard the containership "Shanghai Express" of Hapag Lloyd Container Linie GmbH.. Specifically, the ACME ECU was set up to operate under a proper switching scheme with the existing main engine governor of the ship, which, for reasons of operational safety, was not disconnected. The ACME TP System provided the ACME ECU with load demand prediction, enabling for safe engine running near-MCR under heavy weather conditions.
The key issue that lead to the successful onboard integration of both ACME systems was the detailed planning of installation works.
Two series of sea-trials of the modified engine control installation for the "Shanghai Express" resulted in class confirmation. Consequently, the over-six-months testing period allowed for full-scale, shipboard validation of the methodologies proposed in the theoretical part of the project, as well as, of the prototype electronic systems developed by the project partners. A side aspect, but yet quite important, was the collection of a large volume (over 3 GB in compressed binary format) of vessel powerplant data series. The data acquired included important powerplant operating variables such as fuel index, requested and actual rpm, shaft torque, boost pressure, as well as, aft ship acceleration.
2.2. Control Problems and Strategies
Control problems, occurring in the transient off-design propulsion engine operation, under heavy weather, were presented in the previous section and are also examined in references [9][10]. In the present section, a more detailed analysis of control schemes, which were used onboard to improve governor performance, is given. All algorithms presented are independent and can be used as add-ons to the conventional PI speed regulator [11]. For the simulations performed, the gains of the PI speed regulator have been adjusted in order to achieve a closed loop phase margin of approximately 60 degrees. Each algorithm is briefly described and simulation runs with and without that specific add-on are presented.
The overspeed problem during fast load drops
Speed protection must be present when running at high speed during dynamic load variation [12][13][14]. It is very important that the protection algorithm acts rapidly in order to minimize the maximum speed and the necessary fuel index reduction [15][16].
When the engine exceeds a predefined protection limit, for instance the rated MCR speed, the adopted strategy is to reduce the integrator contribution to the PI controller with a rate proportional to the crankshaft acceleration, but only as long as the acceleration is positive. In order to make the protection effect rather independent of plant specifics, the reduction rate is made proportional to plant inertia and maximum torque. In cases of very fast load reduction a tighter protection is necessary. This protection is made active after a second higher speed limit is exceeded, for instance 105% of MCR speed. It reduces the integrator, not only proportionally to the acceleration, but additionally proportionally to the distance between actual speed and the first limit. A suitable acceleration signal is available from a specifically developed adaptive crankshaft rpm filter. The filtered signal does carry some noise and it is therefore not suited for direct use in a control loop. The signal is however perfectly suited for the purpose described here. Figure 1 shows the simulated response during a test scenario based on propeller torque data obtained from model basin HSVA data series 241 for model tests in irregular waves (Kq is the relative propeller torque coefficient) [17][18][19].
The described protection is very gentle, but highly efficient, Notice that, during the initial three load drops, the fuel index is reduced less when the algorithm is used. However, the maximum speed still exceeds MCR by 50010 compared to the normal case. The same quality of overspeed protection is provided for the last two extreme load drops but here a larger fuel index reduction is required.
Of course the ship's safety system is still there to capture special situations which could not be detected by the governor.
It will in that case order a shutdown if the speed exceeds 107% of MCR.
