Printer Friendly

All aboard.

Testing the oil of diesel locomotives gives operators valuable information about engine operating conditions. Oil analysis is a tool for evaluating diesel engines for telltale signs of bearing and ring wear, heat degradation, and dirt contamination.

However, the process of obtaining an oil sample is both costly and time-consuming. Typically, locomotives are taken into a service center, where an oil sample is drawn manually from the engine, put into a vial, and sent off to a laboratory for evaluation. Results can take days, or even weeks. Analyses are normally performed on locomotive engines every 92 days, during regularly scheduled maintenance. Many railroads take oil samples far more frequently - as often as each time they fuel the locomotive, which can be every few days.

An alternative system developed by Pacific Northwest National Laboratory of Richland, Wash., promises to give railroads a window on engine performance in near real time, and eliminate the need to obtain oil samples.

Pacific Northwest Lab has licensed the on-board oil analysis system, OilPro (for on-board intelligent lubrication prognostics), to Livingston Rebuild Center Inc., a locomotive repair, rebuilding, and leasing facility headquartered in Livingston, Mont. The company plans to incorporate the OilPro in a service to its customers beginning early next year. LRC will manufacture OilPro devices at its research and development facility, in Richland, Wash.


As its name suggests, OilPro will have predictive capability, offering railroads the opportunity to improve maintenance scheduling, according to Kingdon Gould, LRC's project manager. The service will provide diagnostic oil analysis that is currently supplied by off-site laboratories. In addition, the system's prognostic capability will allow for improved maintenance scheduling, he said. Gould believes the prognostic capability can help prevent catastrophic failures and extend the service life of the big diesels by 20 to 25 percent.

Bary Wilson, Pacific Northwest's project manager for OilPro, said the system will provide the same data as an off-site laboratory on demand, while the train is en route, and will offer remote access to results. "The goal is to schedule, maybe months ahead of time, the kind of maintenance needed on these engines to keep them running as much as theoretically possible" Wilson said. "If we can detect a $2,000 repair that must be done, and so avoid a $20,000 repair down the line, that's what we want to do, and that's what the overall oil analysis is aimed at."

The OilPro will be capable of checking for several key operating conditions on big diesel locomotives, noted Wilson. One condition is water in the lubrication oil, which can result from a rupture in the heat exchanger that cools the engine oil. "If a rupture occurs in the heat exchanger, there is a chance of getting water in the oil at a fairly fast rate" Wilson said. "That's a main concern early on."

Other conditions include oil viscosity, which can indicate shear stress and heat degradation; total base number and oxidation; traces of wear metals, such as chromium, copper, and lead, which can indicate the beginnings of ring and bearing wear; and the presence of particulates, which can signal the impending failure of engine components.

Gould expects OilPro to give readings that are essentially as accurate as those performed by off-site laboratories, although the information will be derived by different equipment and, in some cases, will be inferred by software. He added that a major benefit is that sample can be taken with much greater frequency, resulting in a better ability to detect trends from the information for predictive maintenance. Information from samples can be processed and received by the customer, probably within an hour, he said.


The OilPro service will consist of two components: the hardware, located on the locomotive, which will sample and analyze the oil; and a software system that may eventually include artificial neural networks. The software will process the information taken from the on-board monitor, which will consist of four sensors. Information from the sensors will be transmitted, via a commercially available communications system such as cellular phone or satellite, to an information processing center operated by LRC in the Richland area. That information will then be made available to the client railroad.

The OilPro hardware is an elaborate oil monitoring system external to the engine. The oil that circulates through a diesel locomotive is housed in a 300-gallon tank and is pumped through the engine at rates that may exceed 300 gallons per minute, explained Wilson. In one application after passing through the engine, the heated oil goes to a direct oil-to-water heat exchanger where it is cooled, and then passes through a 3-inch line to a large filter. The oil sample is tapped after it exits the heat exchanger, so that it can be checked for the presence of water, and before entering the filter, which will remove impurities.


The OilPro system includes a viscometer, which measures the oil's viscosity. A decrease in viscosity often indicates dilution of the oil by fuel or by water, while an increase in viscosity may result from tars and resins, an indication of increased temperatures that could result from shear. The box also performs elemental analysis using X-ray fluorescent technology to measure wear metals. It also can detect certain salts, such as sodium tetraborate, as evidence of water in the oil that may have flashed off from the heat of the engine, said Wilson.

The analytical hardware is complemented by the software component, which will be able to correlate engine faults with specific oil conditions. LRC is currently gathering data representing experience that LRC has gained from a large number of oil analyses linked to actual engine faults. The software will analyze what the oil looked like leading up to a particular failure, and use that information to predict engine conditions early, explained Wilson.

The OilPro concept is the outgrowth of a related research and development project for the U.S. Army at Pacific Northwest National Laboratory, called Tedann, for "turbine engine diagnostics using artificial neural networks." The Tedann technology uses diagnostic networks and model-based algorithms to predict failures and abnormal operations in the M1 Abrams battle tank's turbine engine.

"For quite a few years, LRC has been working on a program to develop a locomotive health monitoring system," explained Gould. "So when we saw that Pacific Northwest Laboratory was working on a similar idea for the tanks, we decided to take a look at it."

Although much of the work the lab was doing with the tanks was not particularly applicable to locomotives, LRC saw that one of the subsets of the Tedann program, which was very basic oil analysis, would certainly apply. "That was the point when we took that seed and grew it into the OilPro program," Gould said.


In the interest of avoiding information overload, the OilPro will flag certain predetermined diagnostic limits, Gould said. In the same way, railroads will be alerted to predicted engine conditions that may require maintenance.

"We don't want to flood them with new analysis every time," Gould said. "We will process the analysis to derive the value-added information from that." He added that OilPro will also provide better consistency in its evaluation of the analysis. "If a railroad has different people looking at data from the same engine over time, they may not recognize a particular pattern of faults that a given locomotive has shown. The OilPro will keep track of trends and do that work for them."

By correlating oil samples with known engine failures, the software should be able to provide prognostic information, said Gould. He expects that capability to increase over time after OilPro has been installed. based on the sampling information they receive, mechanical departments of the railroads should be better able to predict engine problems and schedule maintenance. "For example, they might get back a sample that suggests there is a sharp increase in the amount of lead in the oil, which would indicate that a certain set of bearings is beginning to wear more quickly," he said.

Pacific Northwest Laboratory's Wilson believes that OilPro can be expanded beyond oil sensing to monitor the overall health of the engine. Possibilities include looking at the electrical system and exhaust stack gases, and potentially vibration monitoring. "It's just a matter of adding a few more sensors," he said.

Wilson also suggested that the system has potential applications outside the railroad industry, including its use on ships, trucks, and aircraft. "Any place that you have a large, expensive engine, these things are going to be very cost effective," he said.

RELATED ARTICLE: Combating Engine Failures in Tanks

The Tedann technology for on-board oil analysis on locomotive engines is an outgrowth of another R&D project at Pacific Northwest National Laboratory. Turbine engine diagnostics using artificial neural networks, or Tedann, was initially developed to monitor engine conditions on the AGT 1500 gas turbine engine that powers the U.S. Army's M1 Abrams battle tank.

The goal was to extend the life of the tank, enhance readiness, and reduce the cost of maintenance, said Frank L. Greitzer, who is Tedann's project manager at the laboratory.

Operational prototypes are currently installed on several tanks at the U.S. Army's Yuma Proving Ground in Arizona and the Washington National Guard at Yakima Training Center in Washington state.

The project, which started in 1993, is funded by the U.S. Army Logistics integration Agency and Department of Defense. Initially, Tedann was developed strictly as a demonstration of the use of artificial neural networks to diagnose engine conditions on the M1 tank's turbine engine. This laboratory-based proof-of-concept demonstration was limited to the electromechanical fuel system of the engine, and results looked promising.

Next, the Logistics Integration Agency asked the lab to build an operational prototype that collects and analyzes sensor data on board the tank in real time, and performs not only engine diagnostics, but also prognostics. In funding the prototype, the agency's goal is to achieve and demonstrate prognostic capability. "The main driving force is whether you can predict problems, rather than just identify them once they have been detected," Greitzer said.

The initial phase of the work involved a study of the engine's history, by interviewing experts, looking up records, and referring to engineering reports to gain insight into maintenance issues with these engines.

"We identified a number of faults and conditions that are important, and particularly ones that could be very costly," Greitzer said. Researchers also looked at what kinds of data are required to diagnose and predict engine conditions.

Although some of the data could be retrieved from sensor ports installed by the OEM, the team identified other areas that would require the installation of additional sensors. The present version of Tedann being field tested uses 32 built-in sensors, plus 16 more that have been added using a wiring harness. Many of the sensors measure temperatures and pressures. Chip detectors register accumulations of debris such as fine particles of metal, while a monitor measures vibration.

All the Tedann electronics and sensors are external to the tank's operation, and testers are not permitted to induce faults on the engine to help identify characteristics under certain operating conditions, Greitzer said. To get around that restriction, the project is using simulated data to help identify characteristics of engine faults and to augment data collected from the field. The team is focusing its analysis on about 40 faults or conditions that can cause degradation in the engine's power efficiency. The lab is building a database of information downloaded from the sensors, as well as simulation data of the thermodynamics and physics of how the engines are supposed to work.

The neural networks are integrating a large amount of information from multiple sources to form a picture of the engine and characterize how it's performing, Greitzer said. Data coming in from the sensors may go through rule-based analyses, which will tell if they are above or below a certain threshold. Analyses from neural networks will produce outputs indicating whether the engine condition is normal or degraded. In addition, values will be noted according to trends over time, making it possible to predict how many operating hours are left before a parameter crosses a threshold that may indicate part failure.

Greitzer said that, if the Army adopts the Tedann technology, it can be built into future tanks, allowing mechanics and logistics planners to access the results of the analyses as well as more detailed data than is currently available.
COPYRIGHT 1999 American Society of Mechanical Engineers
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:new system for testing locomotive engine oil; includes related article on preventing engine failures in military tanks
Comment:Pacific Northwest National Laboratory has developed a new system of testing the engine oils of locomotives that promises to provide results in real-time without the need for obtaining engine oil samples.
Author:DeGaspari, John
Publication:Mechanical Engineering-CIME
Geographic Code:1USA
Date:Aug 1, 1999
Previous Article:Trash, HEAT, and ash.
Next Article:The sub that vanished.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters