Introduction
Currently ship owners and operators are facing a difficult situation – especially since the beginning of the economical[ds_preview] crisis. Due to high fuel prices and low cargo rates the vessel operation costs have to be reduced to a minimum. Further on stringent emission reductions are coming up at the horizon.
This situation leads directly to an increasing interest in a continuous optimization of the vessel performance, combined with a deeper knowledge about the engine behaviour. A very good basis for this attempt is a daily sampling of extensive operation data of the ship owner over a long time period [1]. But online measurement data combined with expert algorithms and intelligent trend analysis enables more distinctive possibilities for performance optimization.
Therefore AVL decided to focus the in-house competencies in engine engineering, measurement technologies and process simulation into the OEM-independent condition monitoring system AVL EPOSTM.
System Concept
The analysis of failure root causes lead to the necessity of robust and accurate cylinder pressure measurement as basis for calculations, correlations, hints and – prospective – control for optimal engine behavior [2] [3]. A result of complete engine simulations showed improvements of advices through additional measurement values (e. g. from turbocharger and bearings), several temperatures, pressures, etc. For that reason AVL EPOSTM is designed as open diagnosis platform for all kind of internal combustion engines and their auxiliaries. The system allows the systematic integration of third parties subsystems providing additional information as well as is capable to be integrated in superior ship or plant control systems. Regularly it delivers the engine status, detects failures and trends for observed engine components and gives the probable reasons for the detected malfunction as well as hints for the operator personnel to solve the problem.
Based on the inhouse strength in combustion measurement a special long life time combustion sensor family was developed mainly installed with a direct connected amplifier [4]. This concept delivers a fixed interface which is beneficial for easy installation or exchange. The link to the network is carried out by the smart indicating unit (SIU) which collects combined values of two pressure channels with the crank angle information and delivers these values on demand to the automation system network (fig. 1).
The software calculations are executed on a separate PC connected to the network. This system design reduces the wiring effort and the PC is able to collect every necessary value from the network supporting reliable engine condition information.
A dependency matrix (fig. 2) was developed after extensive analysis of engine and system failures and related root causes.
As seen in this fault-symptom matrix, each fault consists of the combination of different symptoms. These symptoms contain AVL’s expert knowledge, simulations and experience gathered throughout the development of the system. Since each fault is defined with its characteristic symptoms in individual severity, the different symptoms are weighted for each fault. Another fact is that small deviations due to tolerances of the engine and the sensors as well as noise in the signal are difficult to predict but have to be taken into consideration by calculating the fault probability. Therefore each symptom of a fault is equipped with a characteristic rating function composed of cloudy or fuzzy and sometimes load dependent edges. Basically, each symptom of a fault contains two different sets of rating functions, one to calculate the fault probability and one to calculate its severity, which can be regarded as an emergency indicator. The symptoms itself are then combined based on their assigned weights. Below example shall illustrate the main idea of the fault probability calculation process.
Fig. 3 shows the process of a fault probability calculation on an exemplary fault with three symptoms, of which the deviation to a reference value is denoted by S1 to S4.
The evaluation of the cylinder status is done similar to the fault probability calculation. The different faults are combined based on their impact severity on the cylinder condition and their previously calculated probability. This process assigns to each cylinder an individual status. The condition of the engine is then defined based on the conditions of the cylinders and its subsystems.
GUI-Concept
During the development special attention was turned to a simple graphical user interface (GUI) based on traffic-light-style. Herefrom all important information about engine and cylinder conditions as results from the diagnosis can be gathered. It is also possible to display the data in common diagrams. A user-specific configuration of the GUI is feasible, too.
The GUI has different main functions (fig. 4) which can be configured specifically for several user levels, e. g. for engine technician, chief engineer and superintendency.
The main function »analysis« comprises of proven combustion analysis tools from the AVL software Concerto. Several graphs can be displayed, e. g. cylinder pressure, fuel injection pressure, p-V and heat release curves. Furthermore the results of statistical evaluations can be displayed enabling the user a fast analysis of the engine balancing condition. These evaluations are also used as base for subsequent automatic diagnoses. Trend figures and characteristic plots allow a comparison within the data base as well as to stored reference measurements and thresholds.
The main function »diagnose« evaluates probable engine faults from the before mentioned data. The visualization makes use of easy to understand traffic light symbols with green, yellow and red colour (fig. 5). In addition, grey is used for implausible or missing data. Subsequently the system provides a reduction of a very detailed diagnosis to a cylinder respectively engine condition which allows fleet managers a fast classification and search for engines with operational problems.
AVL EPOSTM V1.1
The basic approach of AVL EPOSTM was already published ([5], [6], [7]). A short concept description can be found beforehand. So in the following this paper concentrates on the specifics of V1.1.
The current system version contains a database for storage and comparison of several vessels and engines. This enables also the easy utilization ashore in the superintendency.
Special attention during development was paid to the visualization of trends. The trends of the sampled data can be displayed either chronological or – more sensible – as characteristic plots versus several basic parameters such as engine load or engine speed (fig. 6). A filter function for the data is used to specify the displayed data in either plot. This filter links the two plots and allows the user to visualize the data chronologically filtered on a selected range of a parameter, such as the mentioned engine load or engine speed. Furthermore each measurement can be selected individually and reloaded if requested. The trend is currently used to outline the time dependent engine behavior. Future development will focus on analyzing the gathered data to define a future engine state and allow a predictive planning of maintenance schedules.
An important issue is the detection of sensor faults respectively implausible signals. The integrated self diagnosis of the system monitors continuously the transmitted signals with respect to plausibility and probable faults. In case of occurrence the affected subsystem diagnosis is suppressed and the related display switches to grey as indication for the user. In case a sensor signal is regarded implausible, AVL EPOSTM discards its data and continues the diagnosis on the remaining cylinders and faults. A diagnosis is therefore still possible, if some of the sensors deliver faulty signals or other reasons for implausibility occur.
Before implementation of fuel injection failures into the expert algorithms a systematic investigation of the effects of a wide variety of failures were executed by using the AVL tool Hydsim at different engine speeds and loads. Therefore a detailed model of a typical fuel injection system was implemented. Fig. 7 shows exemplary the model for a slow-speed two-stroke engine with one pump and two fuel injectors.
Several fuel injection abnormalities are investigated. Typical issues are:
• unsteady fuel injection amount, e. g. due to needle coking or seizure
• influences of FIE system component clearance and leakages (see fig. 8)
• nozzle opening pressure variation (fig. 9)
• sticking plunger (see fig. 10)
• spill shock absorber operation quality
In the following some exemplary results from the simulation are displayed. In combination with practical experiences and real occurring failures the already existing fault-symptom-matrix was supplemented, the fault detection strategy was defined and the respective evaluation algorithms were developed and implemented into the system. Of course, the Hydsim investigation showed also very clear that some practically occurring fuel injection system failures cannot be detected only by measuring the fuel injection pressure. Therefore other monitoring strategies have to be developed.
A further feature of the condition monitoring system is the simple integration of other monitoring systems (torque measurement, bearing monitoring, …). This enables the easy and practical utilization of existing systems from various makers.
Kongsberg vessel performance optimizer concept
The Kongsberg vessel performance optimizer concept provides a set of tools that enables ship owners and operators to manage their vessels in ways that are more economical and ecological beneficial, always in compliance with the existing safety regulations. Some of these tools help in early detection of operational problems and upcoming faults, so the operator can change to predictive maintenance for minimizing downtime.
The tools address specific areas for improvement such as engine and energy optimization, how to operate the vessels speed profile, trim, draft, heading and engine speed versus propeller pitch to optimize the total energy consumption. Each of these parts contributes to a complete vessel performance picture, which forms the basis for meaningful adjustments by the crew.
The vessel performance system is based on Kongsberg’s AutoChief C20 technology and integrates all Kongsberg merchant products and a set of third party applications. In this respect the energy performance analysis system is provided by Marorka. It can cover beside the main and auxiliary engines also boilers and waste heat recovery systems. The engine performance analysis part is done by AVL EPOSTM.
The optimum integration of the different tools can be seen from the following exemplary screenshot of a Kongsberg K-Chief display (fig. 11). This allows the crew an easy access of the provided information without observing various additional screens and displays.
Integration of AVL EPOSTM in Kongsberg Automation System
The integration scheme of AVL EPOSTM in the Kongsberg automation system is shown in fig. 12. All main components of the automation system are connected via a redundant maritime CAN open bus. This includes also the AVL cylinder pressure sensors with the smart indicating units (SIU) and other sensors. Thereby the cabling can be reduced to a minimum. The architecture of this system with distributed processing units (DPU) allows a real time data processing, data reduction and transfer.
AVL EPOSTM is operating on a specific server to which the required sensor data and additional information from the alarm and monitoring system are provided. In return the evaluation and analysis results are sent back to the automation system for display. Alarms can be generated also.
The continuously measured raw data are stored due to a defined strategy according to the specific demands, e. g. normally twice a day and continuously in occurrence of failures or abnormalities. These data can be sent ashore to the fleet management also.
Field experience, references
The long lasting cylinder pressure sensors are an important heart piece of a reliable and sensible condition monitoring system. So the development of such sensors started already ten years ago. In the meantime the sensors have been tested in several engine types on OEM test beds as well as on field installations. Not only the application on various engine types but also with a wide variety of fuels like HFO, gases and biofuels were investigated. The successful results were used as basis for the prototype and test installations of the complete AVL EPOSTM system. An important issue is also the implemented leak-proof concept avoiding any leakage in case of sensor damage.
Meanwhile two marine and one stationary system installations with 35 cylinders monitored in total are in the field (fig. 13). Both marine installations are HFO-driven slow-speed two-stroke engines (MAN 7S60 MC-C, MAN 8K98 ME6). The stationary plant with medium-speed dual-fuel engines (Pielstick 18 PC 2-5V DFC) uses mainly natural gas as fuel.
On the two-stroke propulsion engines the cylinder pressure sensors are mounted with adapters before the indicator valves. The stationary engine is equipped with sensors which are directly mounted in the combustion chamber.
The operation load profile of the monitored engines differs a lot. The stationary engine is operated mainly near to full load whereas the propulsion engines are currently often slow steaming with regular speed and load increases according to the operating demands.
The operational behaviour of the permanently installed sensors could be regularly monitored with an additionally available movable sensor. This sensor is also foreseen for the monitoring of the auxiliary engines.
The runtimes of the first projects installed mid of 2008 are approximately 8,000 hours (status end of 2009). These installations have contributed to the development of a new version of AVL’s monitoring sensors launched end of 2009.
All hard- and software installations operate very stable and reliable. Only some minor problems occurred which are not related to the general operation of the systems. The sensors show no tendency to fouling. The sensor adapters are easy to install for the crew. The specially developed graphical user interface is very much welcomed by the superintendents and the board staff.
Experience shows that the design of the adapter for the sensor installation is of high importance for a low contamination and a reasonable temperature level. Further on an easy mounting of the sensors should be achieved. On the other hand the sensors and cables have to be protected against mechanical damages e. g. due to maintenance works on nearby located engine components.
From the early beginning of the installation several interesting issues were detected enabling the operators a deeper knowledge about their engine behaviour.
E. g. during low load operation of one main engine it was detected that the load distribution of the cylinders is unexpectedly unbalanced (see fig. 14) in comparison to higher loads. The detection of this behaviour enabled the ship owner to initiate respective measures together with the OEM on short notice.
Current development activities
The AVL EPOSTM system is under continuous improvement and extension. This includes the implementation of further know-how and system modules from other AVL tools, e. g. from the FEM tool Excite Designer, as well as of additional expert skills. Further on an upgrading of the sensor signal plausibility check is going to be integrated. Another topic is a further refinement of the algorithms.
For the next version several extensions are currently under development. The main items are:
• automatic TDC correction by parallel simulation model usage
• improved hardware self diagnosis by regular monitoring of the sensor calibration
• thermodynamical and vibrational diagnosis of turbocharger behaviour
• implementation of online NOx modelling as well as SOx and CO2 calculation
• trend evaluation with prediction of engine component behaviour
In the future it is foreseen to cover in several steps the monitoring of the complete engine room with the engines and also the concerned auxiliary systems.
Conclusions
Since 2009 AVL EPOSTM is available as V1.1 on the market. This version concentrates mainly on the permanent expert analysis of the combustion process and the fuel injection. The integration of a capable data base allows the utilization not only on the vessel but also in the fleet management ashore. The characteristic trend plots enable a fast visual judgement of possible changes in the operational behaviour of the monitored engine. A high analysis quality is achieved by the integrated self diagnosis
The system can be utilized on all types of large-bore engines for propulsion, auxiliary purpose and land-based power generation. The actual main focus concentrates on marine Diesel engines but also biofueled engines or gas engines can be monitored. For the latter it is foreseen to provide a system version considering additional specifics of gas engine operation in a while.
Though AVL EPOSTM can act as stand-alone system, the optimum performance is gained fully integrated in Kongsberg’s vessel performance optimizer for marine applications which provides all kind of additional data from the alarm and monitoring system as well as displays the results with easy access for the users. Vessel operators state that they appreciate this delivery of monitoring information independent from the engine makers very much.
The trouble-free acting prototype and test installations prove the reliability and accuracy of the system for a long time period. The gained experience allows a fine tuning of the analysis software resulting in a further improvement of the product.
The system enables the ship operators to gain a detailed knowledge of the engine behaviour and to optimize the engine operation for minimized fuel consumption as well as for minimized emissions. This is valid for all engine load ranges. So this system is also a perfect tool for the monitoring of the currently often demanded slow steaming with approximately 40 % engine load or even super slow steaming with engine loads down to 10 %.
The next steps for further development and system extensions are already defined. So it can be expected that the system will be extended with the before mentioned features in the near future.
Acknowledgements
We thank the vessel respectively plant owners and operators of the prototype and test installations for their contribution, the fruitful discussions and hints for a continuous optimization of the system. These are:
• Höegh Autoliners / Norway
• VASA Kraftwerkepool / Germany
• Reederei Stefan Patjens / Germany
Hinrich Mohr, Norbert Mayrhofer, Christoph Pfister, Rüdiger Teichmann, Rune Johansen
