Cost reduction through online condition monitoring

1. Introduction
Reasons for machine breakdown do have many facets. Besides mechanical fatigue and overload a main issue is the lubrication. The most known crucial factor is the particle contamination. Particle contamination can result from extrinsic factors, such as particles passing through the breathers and sealings, or intrinsic, e.g. wear that is caused by mechanical load or secondary wearout.

Other causes for improper lubrication are e.g. water ingression, lack of lubricant, oil deterioration or the use of a wrong lubricant. For roller bearings numbers are given, that up to 80% of all breakdowns are caused by these influences [1]. See Figure 1.1.

2. Online oil condition monitoring
The term condition monitoring generally refers to the determination and interpretation of status information for machines, systems and their components with the objective of providing condition-based, predictive maintenance. This allows for the early detection of incipient damages and for the performance of condition based maintenance work.

The significant advantage of online analysis compared to laboratory oil testing is the continuous evaluation of oil condition parameters. Thus, online analysis is based not just on a random snapshot, but considers the development of the oil condition over time.

Condition monitoring allows a recognizable monetary advantage by optimizing the maintenance strategy. Figure 2.1 shows the different maintenance strategies.

In a reactive maintenance strategy a service will only be performed if an error occurs, what leads to high costs through damages and machine downtime. The preventive maintenance tries to reduce the costs by examining and changing parts in a fixed service interval. However, real stress and wear of the machine lead to the fact that parts are either exchanged too soon or too late. Continuous monitoring of the machine condition overcomes this disadvantage by allowing a predictive or condition based maintenance strategy.

In Figure 2.2 the condition based maintenance strategy is displayed. The sensor data and machine operation is analyzed and used to show the current machine condition. By automatically evaluating the data with smart algorithms and set limits, a pre-processing is made to generate condition reports. These reports are used to predict condition changes and plan service tasks.

Today online oil condition analysis mainly focuses on measuring pressure, temperature or flow. However, these parameters do not take all relevant factors into account. To monitor more relevant oil parameters various types of sensors with different measurement principles have been developed. Detectable critical conditions are for instant excessive wear, contamination with solids or fluids, humidity, leakage, usage of wrong oil etc. Additionally modern oil condition sensors are often multi-parameter sensors and can therefore replace sensors already used in machines, such as temperature, level or flow measurement sensors. This reduces the cabling and integration costs significantly. Furthermore these sensors enable a better documentation of machine service and maintenance and minimize costs by elongating maintenance periods and reducing the downtime of the machine. Therefore the main reason to use condition monitoring is economic. See Figure 2.3 and 2.4.

3. Application examples
The economic advantage of condition monitoring is related to oil cost savings, reduction of damages and serious consequential damages, avoiding downtime and optimal planning of maintenance work and repairs. Savings stemming from the reduction or avoidance of machine and system damages are difficult to calculate as a lump sum over the lifetime of a machine and can only be estimated. On the other hand, savings associated with longer intervals between oil changes and oil analyses are easy to calculate and the return on investment for the use of a condition monitoring system can be determinated.

In this section different application examples are given to demonstrate the potential of oil condition monitoring.

3.1. Monitoring of oil aging
The example displays the measurement results of a multi-disk braking system with a robust design for high load use. The brakes are equipped with a cooler to be suitable for every environment. The condition monitoring sensors have been installed in the cooling oil tank and connected via CANopen to the PLC of the machine. The PLC integrates the sensor data into the user interface and displays the status to the operator [3].

Initially the service interval and lifetime for the braking cooling oil, a low viscosity mineral oil with high additive level, has been set to 2000 hours by the machine manufacturer. Since the machine is strongly used the oil of the breaks needed to be changed up to four times a year. The actual oil lifetime was estimated to be much longer on an average machine. However, to be on the safe side and avoid oil checks between the services, the manufacturer of the machine has adapted the service interval to the worst case scenario. The task was to extend the oil change interval and detect condition changes.

The used sensor is able to measure multiple parameters of the oil and compensate the parameters for various disturbing influences. The sensor stores the initial fresh oil parameters in a reference memory which are continuously compared with current parameters and the thresholds of allowed deviation. Furthermore the current oil parameters are stored and used to estimate the change rate and calculate the expected remaining lifetime of the oil with integrated models and algorithms. Figure 3.1 shows a schema of the algorithm.

Each individual calculation leads to a predicted remaining useful lifetime (RUL) of the oil, which is in consequence benchmarked and checked for plausibility. In the end the user gets a RUL calculation based on the thermal load, measured chemical changes of the oil and parameters set by user or constructor of the machine.

Figure 3.2 displays an example of a measurement in the two independent braking cooling oil circuits at one machine. The chart shows the temperature compensated relative permittivity and the allowed thresholds for the given configuration during one oil life time period. One can see that the initial fresh oil value is slightly different between the two sensors. That can be explained by the separated tanks that have a different amount of old oil remaining at the beginning. The level of relative permittivity is increasing over time, which indicates rising TAN due to oil aging.

The relative permittivity of a fluid describes its polarity. As the polarity of oil is subject to changes caused by deterioration, the aging process can be monitored by means of the relative permittivity. Furthermore it is also possible to discriminate different base oils according to their permittivity value. This also allows the detection of oil mixture, improper type of oil, oil refresh and oil change [2]. See Figure 3.3.

To justify the significance of the measured parameters theoretical considerations and real world tests have been made. In Figure 3.3 a comparison of rel. permittivity signal with the laboratory measured TAN is shown. It can be seen that there is a good correlation between the two values, what can be explained by the higher polarity of acids. Therefore the permittivity is a good indicator for oil aging.

Figure 3.4 displays the measurement of the temperature compensated conductivity and corresponding thresholds in the same application. Once again the initial value slightly differs but has the same trend.

The interpretation of the conductivity on the other hand is more complex. Depending on the initial level it gives either feedback about the additives or chemical aging products in the oil. As a high start value of conductivity is decreasing steadily, the process can be regarded as a degradation of additives in the oil. The initial value does also give an indication of the additive level of the oil as can be seen in Figure 3.5.

The evaluation of the braking system condition has been performed for about 4500 hours. During this period regular oil samples have been drawn and analyzed in an independent laboratory. As a result the sensor calculates the so called Aging Progress (AP). This parameter expresses in percentage how much of the maximum lifetime of the oil is consumed. Figure 3.6 shows the calculated aging progress over time for both oil circuits, overlaid with the conclusion of the laboratory results that are numbered within the diagram.

The conclusions of the laboratory analyses have been as follows:

1. At 0 hours:
   Fresh oil, no rating, values stored as reference values.
2. At 1.240 hours:
   Viscosity dropped, iron increased, other parameters ok. We suggest normal inspection intervals.
3. At 2.360 hours:
   Iron increased further due to corrosion or wear, viscosity dropped again. We suggest monitoring the oil with the next oil analysis.
4. At 3.400 hours:
   Iron increased further due to corrosion or wear, viscosity stays at lower level than the reference sample. We suggest monitoring the oil with the next oil analysis.
5. At 4.600 hours:
   Iron increased further due to corrosion or wear, viscosity dropped again and shows a stable trend, oil is contaminated with particles. We suggest monitoring the oil with the next oil analysis.

Given the former change interval of 2.000h, it was thereby possible to extend the change interval of the oil significantly and thus cut the maintenance costs.

3.2. Water contamination
High water content is also known for a negative impact on oil lifetime and therefore on machine reliability. Especially free water causes a higher corrosion rate of metal parts, leads to hydrolysis and therefore accelerates oil degradation and other aging effects. Free water can also decrease load capacity of the oil and leads to cavitation within pumps and valves. Typical sources of water ingress are shown in Figure 3.7 [4].

The measured online value is the relative humidity (RH). Relative humidity φ is the ratio of solved amount water in the oil ρW to maximum possible amount of water at the saturation point ρW,max.

The saturation point is highly dependent on temperature and on the oil type. Therefore the relative humidity is changing if the temperature of the oil is changing, even though the absolute amount of water may stay constant. In general oils can solve a higher amount of water when the temperature is rising. To demonstrate this temperature dependence, the relative humidity value has been measured at different temperatures while keeping the absolute water content constant. The results for three different oils can be seen in Figure 3.8.

In contrast Figure 3.9 shows how the relation of relative and absolute humidity at constant temperature does look like.

One can see that the relation depends on the oil type and that in general, mineral oils have a lover saturation point than ester oils at same temperature.

To demonstrate the benefits of the humidity monitoring the measurement of the braking system described in the former example has been evaluated for two machines of the same type. One of these encountered severe problems due to a fault of a service technician. During a scheduled service the cooling oil of the brakes had to be exchanged, but wrong fluid was used instead of the special cooling oil. Even though the cooling oil (mineral oil) has a special color to be distinctive, a cooling fluid for the engine (water/glycol) was filled in. Figure 3.10 illustrates the resulting humidity measurement.

The different numbered sections can be described as follows:

1. Relative humidity level of the oil before fluid change.
2. Sensors notify dangerous high relative humidity level and gives out a warning.
3. The service team reacts, exchanges the fluid with correct oil type and repeating the process to flush the water out of the system.
4. The process is successful and the relative humidity level stabilizes on the same level as before the wrong refill. The braking system has been saved by the early warning.

The humidity warning was directly shown to the operator on the machine control display. Therefore the operator was able to react immediately and prevent damages of the braking system or greater damage through a failure of the system.

Another typical error scenario occurred with the described machines during the bringing into service period. After the production of the machine and its test on the manufacturer site the machine has to be disassembled and transported to the customer’s site, where it is assembled again. During the transportation the machines can suffer bad weather conditions and other hazards of transportation. Therefore the machine has to be inspected after assembly on customer’s site and tested before putting into service.

During the transportation of the machine a sealing failed. Condense water and maybe rain water leaked into the braking system. The installed sensor detected the high water amount in the cooling oil during the putting into service test and gave out an alarm. In consequence the oil could be exchanged by the manufacturer before causing breaking damages.

The described example shows the benefit of online relative humidity measurement. The user of the machine receives direct feedback if contamination occurs. By avoiding damages or downtime and thus increasing customer satisfaction the monitoring of the humidity proved its benefit.

3.3. Dehydration of a turbine
A laboratory analysis gave alarm that the oil of a turbine was highly contaminated with water. Therefore a dehydration unit was installed at the turbine to purify the oil. The rel. humidity of the oil in the suction line of the dehydration unit was measured. See Figure 3.11.

A general example of a monitored dehydration process can be seen in Figure 3.12. Beginning with extreme high water content the absolute humidity is falling steadily. As the oil is fully saturated the rel. humidity value is showing 100% until the saturation limit is reached. Till that point there is free water in the system. Afterwards the high sensitivity of the rel. humidity measurement can be seen and a limit for the purification process can be set easily.

With the dehydration unit it was possible to dry the turbine oil at first, what was double checked by a lab analysis. However, after the turbine was running again the humidity was once again rapidly rising, what can be seen in Figure 3.13.

Surprisingly the latest laboratory analysis taken by the machine operator showed a low water content. Therefore a second analysis was performed with a mobile service unit. At first an oil sample of 10l from the lowest point of the reservoir was drained and analyzed. The result of the humidity measurement shown in Figure 3.14 confirmed that the water content in the sample was very high.

To check the overall status of the oil a second measurement has been made. This time the test point was at the pressure line after the pump. The measurement result can be seen in Figure 3.15.

As the humidity rises when the turbine is switched on a leakage was the source of the water contamination. Measuring and observing the relative humidity serves as an early indicator for harmful free water. Continuous monitoring of the system helps to avoid damages and downtime in the future.

3.4. Wear detection through particle monitoring
During the operation of hydraulics and gearboxes contamination and wear causes high costs for the machine owner. Generally oil filters are used to keep the cleanliness of the system under the recommended limit. However, the machine owner has a sustained interest in monitoring the particle concentration to detect problems at an early stage. Therefore another major application for online oil condition sensors is the cleanliness or wear monitoring.

For particle and wear monitoring various types of sensors with different measurement principles are available. Most common standard today is the optical particle measurement, by means of the light extinction principle according to ISO11500:2008. The particle concentration is expressed e.g. as a cleanliness class according to ISO4406:1999.

To demonstrate the benefit of a continuous particle monitoring the example of a loading machine is presented. The contamination was monitored to detect leakages or wear before a part is failing and thus prevent expensive downtime. The measurement results can be seen in Figure 3.16.

The measurement shows a very clean hydraulic system in normal operating state, which is the result of effective bypass filtration. The cleanliness of the system was monitored continuously and an alarm was set if a programmed limit was exceeded. During the operation suddenly a sharp increase in the particle concentration was detected. An alarm was generated and a service technician was sent to check the machine. During the inspection it was found that a large amount of wear particles had been created in the system by a damaged pump.  After the repair and the subsequent cleaning process, the original low level of contamination was restored.

4. Integration and data management
Nowadays the system integration of sensors is made easy through various interfaces. Furthermore a broad range of accessories is available to access the monitoring data of any machine in real-time. This makes it possible to analyze the condition and generate messages automatically and even predict hazardous conditions before severe damage occurs. 

A particular challenge with accessing machines remotely is reliable communication, especially when the machine is located in a far-off area or is simply mobile and not stationary.  See Figure 4.1.

Modern remote solutions do connect oil condition sensors directly to an internet database with a corresponding user interface. This cloud service makes it possible to view the machine condition directly everywhere and record it for subsequent examination or export. It is also possible to set limits for an alarm handling that automatically generates email or SMS messages. The user can also access the data in the cloud with any recent smartphone, obtaining a quick overview of the machine condition.

For communication with the internet, a gateway can provide an integrated Ethernet interface and an integrated GSM module. The Ethernet interface is generally used for stationary machines with existing internet infrastructure and the GSM connection is generally preferred for mobile machines. An example of the structure of such a remote condition monitoring system is shown in Figure 4.3.

The remote connection of condition sensors can help experts to evaluate the condition of the equipment and its performance without having physical access to it. In the event of an unusual system status, the cause of the error can be contained. This helps to reduce maintenance costs and eliminates errors before they cause greater damage.

Other advantages of the remote system, however, are not obvious at first glance: On-site technicians can receive support from system engineers at central headquarters, reducing travel costs. Remote support can also help reduce the risk and duration of commissioning by monitoring the hydraulic system. In day-to-day business, the system manufacturer can better support the customer through availability of the most up-to-date data and recording compliance with service intervals.

Remote monitoring has further advantages: system supplier development departments can incorporate the telemetry data in their strategy to continuously develop and improve their products. Monitoring the performance data of the equipment improves understanding of their own products and can contribute to the continuous optimization of those products.

5. Summary
Operational reliability and system availability are increasingly important to operators when it comes to purchase decisions. Online oil condition sensors enable new maintenance and service concepts – making it possible to lengthen oil change intervals and minimize production downtimes. These sensors can give out an alarm if the oil needs to be changed, a wrong type of oil is used or detect if the oil has been contaminated with water, other liquids or solids. By reducing the costs for hardware, integration and data analysis OCM systems have become an effective way to reduce the operating costs, which has been proven in various applications.

With the help of remote monitoring, the advantages of oil condition sensor technology can now be implemented even more efficiently as the data from different machines can be accessed both ways, locally and global. Using this information, service tasks can be coordinated and in the event of unexpected results the equipment can be kept safe until maintenance is possible.

List of References:
[1]    Gold, P. W.; Aßmann, C.; Loos, J.: Biologisch schnell abbaubare Schmierstoffe für Wälzlager, Gleitlager und Freiläufe. In Lehrgang Biologisch schnell abbaubare Schmierstoffe, Technische Akademie Esslingen (TAE)
Published: 2000

[2]    Meindorf, T.: Sensoren für die Online-Zustandsüberwachung von Druckmedien und Strategien zur Signalauswertung, Dissertation, Aachen, RWTH, Verlag Mainz
    Published: 2005
    Page 42 – 63

[3]    Dyck, Harry: LubCos Brake Cooling Oil Aging Monitoring, ARGO-HYTOS Application Notes,
www.argo-hytos.com
    Published: 2012

[4]    ARGO-HYTOS: Fluid Management Technical Handbook, The ARGO-HYTOS guide to fluid management and oil condition monitoring
www.argo-hytos.com
    Published: 2009

[5]    Hendrik Karl and Steffen Bots:
“Humidity Saturation Limits of Hydraulic and Lubrication Fluids”, OELCHECK GmbH, www.oelcheck.com
Published: 2012

Dipl.-Ing. Roman Krähling
ARGO-HYTOS GmbH, 76703 Kraichtal-Menzingen, Deutschland

Dipl.-Ing. (FH) Harry Dyck
ARGO-HYTOS GmbH, 76703 Kraichtal-Menzingen, Deutschland