Amount and size distribution of wear particles in engine lubricating oil are indicators of the current machine condition. A change in size distribution, especially a rise in the amount of larger particles, often indicates a starting wear of some machine parts. Monitoring wear particles contained in lubricating oil during normal machine operation can help to identify the need for maintenance and more important to prevent sudden failure of the machine.
An optical method is used to image a thin layer of oil to count and classify contained particles. Therefore, a continuous flow of undiluted oil from the oil circuit of the machine is pumped through the measurement instrument. Inside the instrument, the oil flow is directed through a thin transparent flow cell. Images are taken using a bright LED flashlight source, a magnification lens, and a digital camera. Algorithms have been developed to process and analyze the images. They are capable of compensating for variations in background brightness, of differentiating solid particles from non-solid particles (such as e.g. air bubbles), of classifying particles and air bubbles by size and shape. Additionally the class growth rate is monitored and classified by a prediction software unit, which is able to identify critical situations.
The image analysis algorithms were trained with different oil samples from engines with varying damages. A differentiation between the causes of damage on the basis of the calculated particle properties was shown. Furthermore the system was successfully tested in various applications at engine test stands. It was capable of indicating an upcoming crankshaft bearing damage during a short test run. Using gas bubble count and size distribution, a value for the dispersed gas concentration could be calculated, which shows a good correlation with dynamic engine behavior.
