TP System functionality and requirements
Specific hardware has been designed and developed for the on-board, real-time implementation of the TP Algorithm (TPA).
The main objective of the on-board, real-time TP System is to provide propeller torque predictions, in order to assist the main propulsion Diesel engine governor (controller). Knowledge of propeller torque profile in advance, allows for advanced control schemes to be used, which in turn allows for smoother engine running and, in general, safer operation.
The implemented system acquires on-line measurements of both aft ship vertical acceleration and shaft torque. The variables recorded by the on-board TP System are the following:
l) Engine boost pressure
2) Engine speed setpoint
3) Propeller shaft torque
4) Propeller shaft speed
5) Aft ship vertical acceleration
6) Requested fuel rack position by the ACME Governor
7) Actual fuel rack position at fuel pump
The TP System has been set up so that it provides the ACME experimental governor with the speed setpoint decrease rate, instead of the predicted propeller torque value after several seconds. This strategy is directly derived from the fact that when engine operating conditions are close to MCR, a large propeller torque sink can cause a great amount of overspeed. Therefore, overspeed can be avoided if prior to the minimum torque occurrence, an appropriate reduction of the speed setpoint is applied.
Under this perspective, the TP System can provide the ACME Governor with the speed setpoint decrease rate. The time interval in which the speed setpoint decrease rate must be applied, is defined by a validation signal. This digital signal is set to logical ' I ' during the interval in which the speed setpoint must be decreased by the rate indicated by the value of the speed setpoint decrease rate.
Consequently, the outputs of the TP System are the following:
l) Speed setpoint decrease rate (analogue)
2) Validation signal (digital)
The embedded TPA software
The embedded version of the TPA is based on the original TPA. An acceleration event occurs when the vertical aft ship acceleration reaches a local maximum whose value exceeds a specific threshold (around 0.5 m/sec2). Such an acceleration event triggers the torque prediction process, which is based on the similarity of the occurring event and the events already stored in a dynamic event database. If an event occurs but cannot be correlated satisfactorily with an already recorded event, then the dynamic database is updated. Therefore adaptively of the dynamic database and the prediction algorithm is ensured.
The speed decrease rate generation algorithm is based on the following equation for the "safe speed setpoint",
Nsafe / MCR = 0.75+0.25 Qmin/QMCR
The TPA also provides the time-to-maximum-under torque (Tut), i.e. the estimated time interval until the minimum torque value occurs. Therefore the speed decrease rate, a, can be calculated according to the formula,
a = (Nord - Nsafe) / TUT, only if Nord ＞ Nsafe
where Nord is the engine speed setpoint (relative to MCR) when the acceleration event occurs.
Speed setpoint decrease rate, a, is delivered as an analogue output of the TP System directly to the ACME Governor. The validation signal is kept high for TUT seconds after an event occurrence.
3. RESULTS
The technical achievements of the ACME project were:
1. A digital ECU, with the standard speed regulating function, additionally implementing advanced control schemes for engine operation in heavy weather. This was primarily developed by MAN B&W.
2. An electronic Torque Prediction System, implementing the real-time Propeller Torque Demand Prediction Algorithm. This was primarily developed by NTUA-LME.
Engine control relies more and more on digital electronic technology and model-based strategies [20]. Therefore, the ACME technical achievements may find extensive use in the development of the new generation "intelligent" engines. For these new marine Diesels, real-time control can be applied on fuel injection timing/ profile and exhaust valve opening, except fuelling that is controlled by the speed governor, as in conventional plants.
Major advances in marine propulsion plant control lies on the cutting-edge of marine engineering. Advanced control strategies, if successful, can lead to a new generation of marine Diesels with lower running costs, environmental compatibility, increased efficiency and higher reliability. Moreover, these achievements are possible without any modifications to the baseline engine design, allowing for modifications and improvements to be applied to existing and operating powerplants.
In that respect, add-on installation of the ACME electronic systems is possible on existing plants for performance amelioration and extension of powerplant capabilities under adverse operating conditions.
4. CONCLUSIONS
4.1. Propulsion plant simulation
In the ACME project it became clear that two types of simulation models can be used in parallel for a new propulsion installation study. The two (2) types of simulation models provide complementary results and insight regarding operational details of the installation. The two (2) types of models are given below:
a) For the investigation of engine and turbocharger matching, as well as, other in-cylinder and performance aspects, detailed thermodynamic models (such as the ones used by the ACME project participants NTUA, MAN B&W and ABB, namely MoTher, TAPCODE and SiSy respectively) of the individual cylinders, turbochargers, scavenging and exhaust receivers are required. This type of models is however computationally rather heavy and requires a wide range of input including engine geometrical/configuration data, as well as, experimentally derived curves or maps. They have been used to provide insight in the overall system behavior and data for developing simpler and computationally lighter models [16].
b) For the development of the governor/electronic control unit, transfer-function models can be used. This type of model includes the behavior of fuel actuator and rack, but does not include the thermodynamic simulation of the cylinder process. This results in a model requiring much less computational effort, enabling effective development of the governor functionality, and allowing to simulate a wide variety of the scenarios recorded in tank tests. Therefore, the use of transfer function models proved to be fully acceptable for the specific purpose of ECU development.
