Vibration Diagnostics of Marine Diesel Engines Malfunctions Connected with Injection Pumps Supported by Modelling

The article presents main causes of malfunctions in marine diesel engines fuel injection systems. The measurement section focuses on damage to injection pumps. Measurements were carried out for three technical conditions of a given cylinder injection pump. At the same time, a simplifi ed vibration model of the engine was presented allowing to indicate sensitive components in the frequency domain enabling identifi cation of damages to injection pumps working with individual cylinders. Finally, the results obtained by modelling were compared with the results recorded during laboratory measurements.


1.INTRODUCTION / Uvod
As far as marine equipment strategy of operation is based on their current technical condition, a set of reliable diagnostic methods should be available. They are used for clear defi ning technical condition of all components important from the point of view of reliability. One of such methods is vibration diagnostics. Use of vibration measurements in diagnostic systems allow to determine the optimal moment of shutting down the machine in order to perform necessary repairs. The moment of defect occurrence and its propagation, up to the catastrophic damage, could be illustrated as on fi gure 1. It should be assumed that the moment of defect appearance do not eliminate the engine from further operation. However personnel have to be aware of the situation and have tools enabling observation of defect development. The main purpose of the marine diesel engine injection system is to provide appropriate fuel doses with correct atomization of fuel in the cylinder at a certain angle before TDP. Each of the mentioned parameters describing the operation of the injection system is characterized by certain limits given by the engine producer. As long as these values do not exceed the permissible (limit) values, the engine is in a full technical effi ciency condition. After exceeding the limits, the condition is being changed to a state of partial technical effi ciency. However, it can still be operated. It is important that the technical staff is aware of the need to quickly repair of developing defect. If an existing defect is not detected at this stage, it usually develops very quickly leading to engine failure.
All working machines, also those in good technical condition, are a source of vibrations as a form of dissipated energy. In the vast majority of cases (but not always, for example, vibrations caused by the fl ow of the medium), they can be linked to periodic phenomena occurring in the machine, e.g., a rotating shaft, gear teeth or frequency of supply voltage in electric machines. Usually, the relationship between values and magnitude and/or phase of their occurrence is relatively simple do determine. From the point of view of vibration diagnostic it is useful and serves to fast assess the technical condition of the tested device.
Taking into consideration many years of experience gathered at the Institute of Construction and Operation of Warships, an analysis of the occurrence of operational incapacity of marine engines operated in Polish navy was carried out - Figure 2. The conducted analysis presented that the most common defects occur in the following functional systems [6]: -engine fuel supply system -72%, -timing system -19%, -engine power supply system -9%. instalacije goriva [6] From the diagnostic point of view, exceeding the permissible values connected with the operation of fuel systems of marine diesel engines has two main consequences: -leads to changes in mean eff ective pressure (Fig. 3), -changes in pressure distribution as a function of crankshaft angle (CA). The paper is focused on the subject of injection pumps vibration diagnostics because although their failures do not constitute the biggest percentage their identifi cation can be considered as the most diffi cult. On small vessels (particularly naval vessels), there are no portable stands for testing injection pumps, unlike stands for injector testing.
Changes in the technical condition of the fuel system could cause a decrease in the quality of the inside cylinder processes. Additionally, in the long term, could cause serious secondary damage to other components of the naval power unit, such as turbochargers, fl exible couplings or torsional dampers. It is the reason why early detection of changes in the technical condition of injection pumps can also prevent secondary failure to other components of the propulsion system. Currently, one of the best methods to accurately indicate the reason of disturbances in inside cylinder pressure is the indication procedure. In many cases, it cannot be made due to the lack or failure of indicator valves. The analysis of the literature gives the information that issues related to the vibration diagnostics of diesel propulsion engines, have not found a clear and acceptable methodology. Despite the inability to indicate, it is advisable for the personnel to have diagnostic information from other sources [7,8,11,12]. In view of the above, an important issue is the selection of measurement and analysis optimal method of vibration signals. It allow to determine the technical condition of injection pumps of the marine engines.

THE COURSE OF RESEARCH / Tijek istraživanja
Gas pressure pulsation in the cylinder will result in a change of force aff ecting its crank-piston mechanism and the surfaces closing the combustion chamber [3,4,10]. Therefore, it can be assumed that the vibration parameters recorded in the immediate vicinity of the cylinder operating with the faulty injection system will diff er from the reference one [4]. At the initial stage of the study to determine the diagnostic sensitivity, the comparison of the amplitude spectra of the vibration accelerations obtained from measurements carried out at various points of the engine was made. On this basis, two measuring points located in two mutually perpendicular directions V (vertical) and H (horizontal) have been selected as the most susceptible to changes in the indicated pressure - Fig. 4. The substantiation is, the distribution of the main forces occurring in the crank-piston mechanism of the working diesel engine, what is presented. After performing the analysis of the suitability of the recorded vibration parameters, a further part Measuring points in the direction V (fi g 5) were located on the tightening bolts of individual cylinder heads. Signals of accelerations were measured which were fi ltered by high pass fi lter with cot of frequency of 0,7 Hz. Subsequently FFT procedure was used. The measurements were carried out in three stages corresponding to various technical states of the injection pump. The simulation of a change in the technical condition of the injection pump was gained by changing the pressure in the injector supply pipe. This change was obtained by adjusting the overfl ow screw on the pump discharge side. In order to determine the eff ect of the position of the overfl ow screw on the amount of fuel injected into the combustion chamber, cylinder pressure was continuously indicated (Fig. 3). During the fi rst stage of the tests, vibration parameters of all cylinders were measured in a state of full technical effi ciency, the results obtained during this stage served as reference. The next step was to take measurements with the injection pumps of individual cylinders switched off , which resulted in the simulation of the misfi ring. During the last phase, the measurements were carried out for partial condition of individual injection pumps. The amount of fuel was reduced to the level at which the indicated pressure reduction was obtained by about 20% compared to the operation of a cylinder with a effi cient injection pump. All stages of the tests were carried out in similar atmospheric conditions and for the same engine torque loads.

MARINE DIESEL ENGINE VIBRATION SIMULATION MODEL / Simulacijski model vibracija brodskog dizelskog motora
At the same time, a simplifi ed model of dynamics was developed in Matlab. The main source of vibration excitations in working engine is rapid pressure changes occurring during the working stroke [1,2,10]. Therefore, when designing the model it was necessary to mathematically describe the course of pressure changes in individual engine strokes. Analyse of literature gives us information about many diff erent diesel engine pressure models, but usually, they are diffi cult to implement in the fi nal engine vibration model [3,7,8,10]. In order to create a simplifi ed pressure model according to the Sabathe theoretical cycle, it is necessary to determine the characteristic parameters of the engine cycle [2,5]: -boost pressure, -exhaust pressure, εcompression ratio, -compression polytrope exponential index, -expansion polytrope exponential index, -pressure increase ratio, ρ -volume increase ratio, r -crank radius, λ -crank radius to the connecting rod length ratio, D -cylinder diameter, ωcrankshaft angular velocity.
To create a model, it was assumed that in the intake stroke (in the range from 0 to 180 o CA) the gas force acting on the piston remains constant and takes the value specifi ed by the formula (1) [2]: (1) where: p 0 -ambient air pressure.
In the next, compression stroke (180 to 360 o CA), the change in gaseous force might be described by the equation In the work stroke (from 360 to 540o CA), the value of gas force will change approximately according to the dependence: The approximate value of the gas force in the exhaust stroke (540 to 720o CA) can be considered as constant and described by the equation: (4) In diesel engines, the initial phase of the expansion stroke is an isobaric increase in volume and in this phase the formula (3) might not be applied. According to the literature [2] this formula can be used. The condition is to replace the gaseous forces greater than those induced by the theoretical maximum pressure p with the forces corresponding to this pressure. The maximum pressure is expressed by: (5) In contrast, the gas forces caused by this pressure are described by the equation: The substitution should occur in the range from 360 to oCA, where is given by the equation: As a result of the implementation of the presented equations and basic dimensions and masses of the Sulzer 6AL 20/24 engine to the MatLab script, the course of cylinder pressure changes shown in Figure 6 was obtained.
Such a chart of pressure is unusable for further create a diesel engine vibration model due to step changes in pressure at the ends of each stroke. Therefore, the authors used one of the nonparametric methods of regression, i.e. weighted localpolynomial regression (loess -locally weighted scatterplot smoothing). As a result of applying the loess function to the course presented in Figure 6, a course much closer to the recorded during the engine's indication was obtained. The comparison of the three real curves obtained from the model and obtained from the model using the loess function was shown in Figure 7.
In order to determine the conformity of the cylinder pressure course obtained by modelling with the course recorded during the engine's indication in the fi rst stage, the Pearson's linear correlation function was used (the dimensions of both matrices are identical). The correlation coeffi cient of two variables describes their linear dependence taking values from -1 to 1. If both variables are N-length then the Pearson's linear correlation coeffi cients take the form: (8) Where and are the mean and standard deviation of the set A, while and are the mean and standard deviation of the set B. The linear correlation coeffi cient can also be calculated using the covariance function: After applying the corrcoef function, a correlation matrix with the following form is obtained: The following values were obtained for the test runs shown on Developing engine dynamics model, the fi ring order was taken into account as well as the distribution of forces in the crank-piston mechanism, also the forces related to the operation of suspended mechanisms were included. The implementation of the obtained pressure model for the engine dynamics simulation model allowed to generate modelled time courses of vibration accelerations for various engine operating states. The comparison of the model chart (in relation to the effi cient engine) and the chart recorded on engine's cylinder no. 1 is shown in Figure 10    In fi gure 13, the green colour line indicates the reference spectrum of engine vibration acceleration; orange colour line refers to the engine working with partially effi cient injection pump on a cylinder which the accelerometers were mounted. Red spectra were recorded after turning off the injection pump. It is observable that switching off the injection pump caused a signifi cant reduction in the amplitude values in the whole measuring range. Similar results were obtained by switching off the injection pumps of the tested engine remaining cylinders. It allows forming the conclusion that ignition misfi res are relatively easily detectable using vibration methods. Which is why in the next part of the work was focused on determining the possibility of detecting the initial stages of the ineffi ciency of injection pumps. Simulated damage to the injection pump consisting in reducing the amount of fuel injected into the combustion chamber caused   Table 1 presents the results obtained from measurements and modelling one of the tested engine cylinders. The results achieved for the other cylinders show the same tendency. Analysis of collected data recorded during operation of the 6AL 20/24 engine with simulated failure of the injection pumps allowed to state that in the case of failure to the injection pump of a given cylinder, consisting in reducing the amount of fuel supplied to the combustion chamber, the vibroactivity of this cylinder is reduced. The same dependencies were obtained by modelling. The developed model requires further tuning to achieve greater adequacy.

CONCLUSIONS / Zaključci
The research results presented in the paper confi rm that there is a strong relationship between the technical condition of marine medium and high speed marine diesel engines injection pumps and vibration parameters recorded on the tightening bolts of engine heads. In addition, it is possible to indicate diagnostically sensitive parameters with simulations. The use of only vibration diagnostics limits the possibility of full identifi cation of changes in the technical condition of injection system. Obtained results will be used to develop the proposed diagnostic method through the use of multi-symptom diagnostics methods. This step will enable identifi cation of the original symptoms of changes in the technical condition of the fuel injection system in a non-invasive manner in the fi eld of all rotational speeds and operational loads of marine diesel engines.