This article will document a complete sampling program at a major industrial manufacturer, the need for full commitment of the user, and the benefits realized when the customer/user understands oil analysis and is fully engaged in the program. A case history will document the decision for a preemptive replacement of a critical gearbox and the resulting cost savings to the customer.
Background
Manufacturing plants with world-class, condition-based maintenance (CBM) or predictive maintenance (PdM) programs use a combination of technologies to determine the mechanical and lubricant health of their equipment. The three most common techniques currently in use are thermography, vibration analysis and oil analysis. When used in combination, they provide maintenance professionals with the information needed to make accurate and informed decisions.
In many cases, equipment critical to plant operations is often "unspared," with no backup unit to replace it when it is not in service. Unspared critical equipment in plants typically have common characteristics:
- They require very high capital investment and are expensive to maintain and repair.
- They are engineered for long service lives when operated within design specifications and in a predictable environment.
- Many are quite large and are made up of several individual components.
- Downtime is quite expensive since production is usually halted when unexpected problems or a system failure is experienced.
Major repairs and overhaul of critical equipment often require a complete plant shutdown, substantial manpower and subsequent loss of production activities. As a result, maintenance managers work to maximize trouble-free equipment operation and ensure repairs are scheduled before a loss of service occurs. Unfortunately, when deciding to remove or repair problem machines from service, the approval process is often difficult. It is not uncommon when engineering and maintenance personnel have evidence of pending problems that a number of meetings must be held with operations management before final action begins.
Of the three primary PdM technologies, oil analysis provides the most information for equipment "end-of-life" decisions and root cause determination. Later in this article, we'll discuss a case study where oil analysis provided the first alert to an impending problem with the findings subsequently confirmed through vibration analysis.
Critical plant equipment requires testing that is broader in scope than the general oil analysis programs provided by oil companies and many laboratories. Appropriate testing should not be derived from a "standard" slate or suite of tests based on a selling price or sample bottle size. Rather, the testing needs to be appropriate to the machine and its application.
Testing should be sufficient to provide an early alert to changes in unit wear, lubricant condition and degradation, and potential sources of contamination to ensure the information needed for an in-depth evaluation is available when problems are indicated. Appropriate testing includes, but is not limited to:
- Wear metal analysis;
- Moisture content;
- Viscosity;
- DR ferrography or PQ Index;
- Acid number;
- Analytical ferrography;
- Particle counting;
- Examination of filter media debris.
Oil analysis is most effective when samples are analyzed on a regular basis, allowing a trend analysis to be developed. Also, results are monitored and measured against not only accepted limits and ranges, but more importantly, against the norms that are established for each particular machine. Accurate trended data permits an experienced laboratory analyst, or the end user, to diagnose and closely monitor the causes and effects of changes within the system.
DuPont, a Fortune 100 chemicals and products manufacturing company, owns and operates over 75 plants worldwide. DuPont USA utilizes all three previously mentioned PdM technologies as part of its corporate predictive maintenance program.
Case Study - Extruder Gearbox Failure Averted
DuPont operates a critical extruder gearbox at one of its chemical manufacturing facilities in the Texas Gulf Coast. The gearbox has an 8000 HP design with a 4:1 service factor - (2000 HP). It has an oil capacity of 2300 liters (600 gallons) and weighs more than 45 tons (100,000 pounds). The unit runs continuously and produces more than 3200 kilos (7000 pounds) of product each hour. The PdM technologies employed at this plant are oil sampling and vibration analysis.
The plant is located in the hurricane zone of the United States. On September 13, 2008, Hurricane Ike made landfall on the Gulf Coast. As part of the company's disaster planning, the plant was shut down in "as is" condition just prior to the arrival of the storm. There was no power or any basic services within the plant during this time and severe flooding due to heavy rainfall and rising sea water from the Gulf of Mexico occurred within the plant.
As part of the company's post-storm disaster recovery plan, all lubricant had been pumped from the unit. It was opened, inspected, wiped clean, then flushed for four to six hours. New oil obtained from storage drums was used to refill the gearbox.
The extruder's lubricant was not contaminated from the storm, however, the drums with the new replacement oil had been stored outside and were heavily contaminated with water. The proactive attempt to prevent problems inadvertently created a problem.
On the first oil sample after the hurricane, Analysts, Inc. laboratory identified .30% water present. Unfortunately, plant personnel were very busy with other problems within the plant and did not react to the report. The next sampling in March 2009 reported .50% water. (See Table 1)
Table 1 - Water and physical property data
DuPont's corporate lubes consultant contacted the plant to discuss the oil analysis results and recommended additional testing of water via the Karl Fischer test, analytical ferrography and a lube filter examination. The Karl Fisher test was to confirm the moisture finding, ferrography was to check for rust and oxides, and the filter exam was to identify collected contaminants.
On May 1, 2009, the ferrogram report revealed moderate rubbing wear (Figure 1), ferrous oxides, a large fatigue particle (Figure 2) and a concentration of spherical particles made of low alloy steel. Antifriction bearings were determined to be the source of the low alloy steel. The spheres and presence of large wear particles were indicators of a potential end-of-life wearing pattern.
Figure 1 - Rubbing wear and spheres
Figure 2 - Large fatigue particle
The lube filter media examination revealed a large quantity of iron oxides, corrosion products, fibers, seal material and varnish. The filter was removed and replaced on May 1, 2009.
As a follow-up after six days of use, the replacement filter was sent to the laboratory for examination. Due to the very short time this filter was in service, there were lesser amounts of contaminants than found in the first filter. This reduction in contaminants allowed for a better evaluation of the wear material. Sliding wear particles and more spheres were identified. Additionally, the laboratory confirmed the continued presence of iron oxides, corrosion products, fibers, seal material and ferrous scale that were previously observed.
Figure 3 - Debris from initial filter inspection
Figure 4 - Debris and large wear particles from second filter
From the data collected and discussions with the lube consultant and plant personnel, it was determined the wear was being caused by a lack of lubrication, high moisture and poor filtration.
On June 1, 2009, the lube consultant, plant personnel and repair vendor met to discuss the available options. Due to business conditions, the unit was to be kept in operation. System monitoring using oil sampling and vibration analysis was done at more frequent intervals, and a kidney loop filtration system was installed. In the short term, spectrochemical analysis results showed a reduction in iron wear. However, after three months, chromium began to appear and after six months, both iron and chromium wear rates increased significantly. Another ferrographic analysis was performed and rated as critical due to the presence of heavy concentrations of ferrous rubbing and laminar wear particles, indications of bearing spalling and oxides.
Table 2 - Spectrochemical wear trend data
Based on all the corroborating data, plant operations and maintenance personnel agreed to schedule the proactive removal and repair of the gearbox. Assuming no delays or unexpected problems, this was a 21-day process that impacted plant revenues by $2.5 million (1,909,000 euro).
There was an unexpected discovery that happened two months prior to the removal of the gearbox. The extruder casing was opened so the gear set could be visually inspected. Just before closing the gearbox, the lube oil pump was activated to check the operation of the input pinion gear spray nozzles. It was found that all four nozzles had blockages, two with minor flow restrictions and two completely blocked. The defective nozzles were providing lubrication to the high speed pinion mesh but not to the bearing systems. The nozzles were removed and cleaned than put back into service. The reinstalled nozzles restored proper oil flow to the high speed gear set, but the damage had been done. In the last sample prior to removal of the gearbox, the wear had increased again to its highest level during the life of the gearbox indicating that the unit was entering into an early failure stage.
In addition to the water contamination the gearbox had experienced, it was determined that partial oil starvation caused the increased wear to occur. If the faulty spray nozzles had gone unnoticed, the repair work would have soon been wasted as the wear to the unit would have continued to increase and most likely caused an unexpected failure during production.
Figure 5 - Plugged nozzles with impaired flow
Figure 6 - Flow from nozzles after maintenance
After the repairs, the subsequent oil sampling and vibration analysis confirmed that the problems were corrected and the gearbox continues to operate without any detectable problems.
Summary
This case history is an example of a strong and successful CBM/PdM program that utilizes multiple technologies to provide world-class service. In this situation, oil analysis proved to be the best technology for early warning of the onset of a major problem. It also allowed continual assessment of the machine's condition until repairs were feasible.
The success of the DuPont program is due to the company's commitment to the program and contributions by many departments and personnel. The testing utilized is based upon the various equipment and applications. More in-depth testing is encouraged and supported whenever systemic problems are identified. Decision-making is based on best business practices and practicality.
The decision to remove the extruder gearbox from service and make the necessary repairs was expensive - $2.5 million decisions are not made lightly. However, if the decision had been to continue the unit in operation, it most certainly would have experienced an early endof- life failure. An unplanned for failure in this case could have cost the plant as much as three times the money spent, with many additional weeks of plant shutdown.
The oil analysis program employed by the plant is a very low cost, high return investment that complements the other technologies utilized.
