Accurate, timely field test results are necessary to develop and validate lubricants meeting frequently changing performance requirements. Field tests can also provide valuable information about performance deficiencies (e.g., soot related wear) which are not apparent in laboratory development tests. Since field tests are time intensive and increasingly expensive, it is imperative that the data generated provide meaningful results with reasonable expenditures.
The data generation and analysis process are being constantly improved according to the principles of quality management. Part of the process improvement focuses on accurate, realistic treatment of the data since more variation is typically observed in field tests than in laboratory tests.
One of the most difficult analytical processes occurs with oil consumption data. Historically, when on-highway diesel engines exhibited oil economy on the order of 850-1700 kilometers per liter (500-1000 miles per quart), the omission of an oil addition data point had little effect on cumulative oil consumption because the typical engine consumed approximately 473 liters of oil during a 483,000 kilometer (300,000 mile) test. Today, however, engines operate for 3404-5106 kilometers on a liter of oil, and cumulative oil consumption for a 483,000 kilometer test can be as little as 142 liters. A missed oil addition (or failure to document an oil addition) produces 3-4 times the effect on cumulative oil consumption data.
Procedures used to increase confidence in field test data include:
Normalization of used oil wear metals for oil drain interval and oil consumption variations.
Application of process control chart techniques to eliminate errant oil consumption and wear metals data.
Use of frequency distribution charts to evaluate oil consumption, piston deposits, and wear performance of different test oils.
Profilometer measurement of cylinder walls to investigate wear performance.
This paper will detail application of the above processes to analysis of heavy-duty diesel field test data.