Comparative evaluation of various methodologies to account for the effect of load variation during cylinder pressure measurement of large scale two-stroke diesel engines

A significant number of fault-detection and fault diagnosis methods are based on the use of the measured cylinder pressure to estimate critical engine parameters i.e. cylinder brake power, fuel consumption, compression condition, injection timing etc. But, the results derived from the application of these techniques depend strongly on the quality of data used. A common problem which has been identified by many researchers is load variation during cylinder pressure measurement. This for some applications (marine) can become significant and in some cases makes unusable utilization of cylinder pressure measurement. According to the conventional measurement technique for field applications, cylinder pressure is measured sequentially instead of simultaneously due to issues related mainly to cost, applicability and complexity. Because of this, the operating parameters that are estimated for each cylinder depend on instantaneous engine load. Therefore when an operating problem or a mistuning is identified, the distinction for the actual cause (i.e. if it is attributed to a malfunction, mistuning or to engine load variation during measurement), is difficult because cylinders are not measured simultaneously. For this reason, in the present work, three methodologies that have been developed to account for the effect of load variation on diagnosis results are presented and evaluated in an attempt to be offered an alternative against simultaneous cylinder pressure measurement. For this purpose, a well validated diagnostic technique, developed by the present research group, is employed and modified accordingly. The aforementioned methodologies have been applied on a large-scale two-stroke diesel engine used for power generation in a Greek island at two different operating conditions. From the application of each method, diagnosis and tuning results are derived which are then compared against the respective ones obtained from the conventional diagnosis technique which neglects the effect of load variation during measurement. The evaluation is based on the comparison of vital engine performance data i.e. brake power output, cylinder fuel flow rate, peak firing pressure, ignition angle and compression quality. From the comparison of the diagnosis results it is revealed that the three methodologies provide adequate results while the one which is based on the use of two cylinder pressure sensors provides a very competitive alternative against simultaneous cylinder pressure measurement, offering the advantage of simplicity and low cost. Most important it is demonstrated the potential for all three methods to propose the required engine tuning to guarantee uniform cylinder operation despite variation of load during measurement.