Written by Grant Maloy Smith, the data acquisition expert
Machines break, that’s inevitable. Using condition monitoring, engineers can accurately predict when this is likely to happen. In this article we will learn:
- Why do machines fail?
- What types of machine maintenance exist?
- Why is preventing machine failures important?
- What machine condition monitoring really is?
Jump to a section
Why Do Machines Fail?
Machines fail due to different reasons and all failures are not the same. Machinery fails or loses their usefulness when they stop functioning in the way they were designed for.
Figure 1: Machine catastrophic failure
This loss of usefulness is broken down into three main categories:
- surface degradation
- obsolescence and
Surface degradation of machine parts results in the machine’s loss of usefulness in the vast majority of cases and comprises mainly mechanical wear and corrosion.
Why Is Preventing Machinery Failures Important?
Every unexpected stop in production, due to equipment failures, has a significant influence on the company productivity, repair and other costs, revenue, profits, and finally competitiveness. It is estimated that downtime costs automotive manufacturers around 22 thousand USD per minute or 1.4 million USD per hour. This is the reason why facility owners are constantly in search of ways to eliminate failures while keeping maintenance costs at the lowest possible level.
Proper Condition Monitoring helps companies to:
- Drop repair costs
- Reduce maintenance costs
- Increase plant life
- Increase personnel safety
- Increase revenue
- Increase profitability
Figure 2: Machine Condition Monitoring Benefits
Types of Machinery Maintenance
Several different types of machinery maintenance exist but in general, we divide them into two categories:
- Preventive maintenance
- Corrective maintenance
What Is Corrective Maintenance?
Corrective maintenance is a technique where correction is done after the failure occurs. It is used for machines that are very low cost, easy to replace and their malfunctioning does not significantly affect productivity.
What Is Preventive Maintenance?
Preventive maintenance is a technique where maintenance is performed before the failure occurs. The most renowned preventive maintenance techniques are:
- Time-based Maintenance (TBM)
- Predictive Maintenance
- Condition Based Maintenance (CBM)
In 1736, when he was advocating for better fire prevention in Philadelphia, Benjamin Franklin famously advocated, “an ounce of prevention is worth a pound of cure.” Surely it is better to prevent a fire than to try to extinguish one. This common-sense approach is at the center of preventative maintenance.
Time-Based Maintenance (TBM)
Time-based maintenance is performed at regular intervals. Refers to the replacement of an item regardless of its real condition. Normally it is performed at intervals prescribed by the machine manufacturer and uses as the basis the Mean Time Between Failure (MTBF) data.
The predictive maintenance goal is to predict when a failure will occur, based on the data obtained from the condition monitoring system. When engineers are alerted that a problem is developing, they can take steps to move the affected system out of a critical position while maintenance is performed. When the problem is resolved they move it back seamlessly afterward. So Predictive maintenance can not exist without Condition Monitoring.
Condition Based Maintenance (CBM)
Condition-based maintenance (CBM) is a maintenance strategy that monitors the actual condition of an asset to deciding what maintenance needs to be done. CBM dictates that maintenance should only be performed when certain indicators show signs of decreasing performance or upcoming failure.
What is Machine Condition Monitoring?
Machine condition monitoring is a process of checking the status of the machinery during its normal operation. It consists of data acquisition, data processing, and data comparison with trends, baseline, and representative data from similar machines.
Machine Condition monitoring with Dewesoft
Machine Condition Monitoring Background
Humans are subject to condition monitoring each time when visiting a doctor for a health check. To know the general state of health of a patient and to follow the evolution of a disease, several examinations are included depending on the age and condition of the patient.
Figure 3: Health-check
Generally, a check-up includes:
- An interrogation: the doctor asks the patient to describe the symptoms he feels.
- Auscultation: Using a stethoscope, the doctor listens to body sounds such as heartbeat, waterborne sounds, and carotid and femoral artery sounds.
- An electrocardiogram or ECG: this examination corresponds to the electrical activity of the heart via electrodes placed at the ankles, the handles, and on the chest.
- X-ray of the chest: it can detect possible problems in the chest, lungs, and heart.
In industry, the condition monitoring of machines is performed by machine doctors called predictive maintenance engineers (PdM's). Their role is to use diagnostic systems to prevent unexpected production downtime and catastrophic failures with minimum production stops and minimum maintenance costs. What parameters do they check? There are plenty of parameters that can be checked ranging from visual inspection, oil levels, oil debris, temperatures, corrosion, vibration, pressures, etc.
The development of condition monitoring started long ago with some very simplified measurements. In late 1850 railway maintenance technicians used wheel-tapping hammers to check the state of the wheels on locomotives. By hitting the wheels and analyzing the sound they were capable of assessing the wheels' state (a wheel with a crack was emitting a dull sound).
Figure 4: Wheel-tapper
Progress in electronics and software development are dramatically changing machine condition monitoring making it simpler to use and way more reliable.
Condition Monitoring Applications
Condition monitoring applies to thousands of applications, but the best known include:
- Industrial plant and facilities of all types: Gearboxes, UPS, AC, electric motors, fans, pumps.
- Pulp & Paper: Blowers, conveyor belts, chippers, chip classifiers, refiners, pressure screens, screw conveyors, agitators, nip monitoring, felt rollers, etc.
- Iron & Steel: Raw material handling machines, conveyor belts, ship unloaders, galvanization plants, stack reclaimers, continuous casters, cranes, rolling mills, annealing machinery, and also pumps, fans, and gearboxes, etc.
- Automotive: Wind Tunnels, Air handling units and pumps in paint shops as well as presses and transfer presses, etc.
- Cement: Crushers, gearboxes, conveyor belts, separators, fans, raw mills, ball mills, elevators, and blowers.
- Power generation plants: gas turbines, steam turbines, water pumps, etc.
Machine Condition Monitoring Steps
To successfully implement a machine monitoring program it is essential to follow a well-structured approach in the following steps:
- STEP 1: Set-up of equipment register
- STEP 2: Assessment of machinery status and their criticality for the facility operation
- STEP 3: Identification of appropriate machine condition monitoring technique for each of the available assets
- STEP 4: Selection of available technologies on the market
- STEP 5: Installation of the condition monitoring system
- STEP 6: Data collection and data interpretation
- STEP 7: Maintenance tasks determination
Let's look at each of the steps in detail.
STEP 1: Set-up of Equipment Register
This step aims to build a register of all production facility assets. The register usually includes:
- Process drawings
- Wiring diagrams
- Exact details of each machine (type, speed, coupling, power, etc.)
- Asset position for easy asset finding
- Unique ID number
STEP 2: Assessment of Machinery Status and Their Criticality for the Facility Operation
Review of past assets failures, analysis of MTBF (Mean time between failure) and MTTR (Mean time to repair), average costs of repair and replacement, cost of downtime, risk of secondary damage should be obtained. This will help us in the identification and selection of the right machine condition monitoring techniques and technologies.
STEP 3: Identification of Appropriate Machine Condition Monitoring Technique for Each of the Available Assets
There are several machine condition monitoring techniques that are used for the assessment of machine condition. Let's take a look at the ones used most frequently.
It has been adopted for machine health assessment over the last decades. There are several methods of temperature monitoring that range from passive, non-contact (using IR cameras) to active sensor-based (using thermocouples and RTDs).
IR scan can give a good overview of machines or control electronics and indicate overheating problems. Contact measurement is very useful for early detection of lubrication related problems but not that much in detecting physical damages such as bearing cracks and spalls.
Figure 5: Temperature measurement using a thermal imaging camera
Check out Dewesoft Temperature Data Loggers
It is a very old and the most commonly used method for assessment of the machine condition. It helps us detect the failure and understand its root cause. Accelerometers are used to monitor changes in amplitude across a wide frequency range. Vibration monitoring permits you to understand phenomena such as misalignment, unbalance, looseness, gear tooth issues, or bearing wear before failure.
Figure 6: Vibration monitoring with a data acquisition system
Acoustic emission sensors are recently being used more and more for condition-based monitoring because of many advantages for the early detection of faults. But it is not a suitable method for permanent monitoring installations because of its problem of massive data storage requirements due to its operation in high frequency (a few kHz to MHz) and high prices compared to other available solutions on the market. It is also difficult to pinpoint the source of the sounds that the sensors measure.
It is a very cost-efficient technology used especially to answer the initial question if a machine is healthy or not. Ultrasound detectors normally measure sound pressure waves in the frequency range between 30 kHz to 40 kHz.
Pressure waves are measured using a resonant sensor which transforms waves into a small electrical charge. It is normally used hand in hand with vibration monitoring techniques. Technicians normally use ultrasound to filter good machines from bad machines and then perform an in-depth vibration analysis of the bad equipment to find the root cause of the issues.
It is normally performed in laboratories using chemical tests to determine the state of the oil. Nowadays tribology sensors for permanent oil quality monitoring exist. The results indicate if the oil should be changed.
This technology is very rarely used to assess the condition of the asset and is more focused on determining the condition of the lubricant (viscosity, basicity, etc.). However, oil level and quality tracking are very important to prevent costly repairs.
Figure 7: Motor oil analysis
STEP 4: Selection of Available Technologies on the Market
As we learned so far, there is a wide range of available techniques for machine health monitoring. The best strategy of every maintenance technician would be to use a combination of all of them to gain the best results. However, due to budget and time limitations, vibration diagnostics in combination with temperature measurements have proven to be the most effective so far.
STEP 5: Installation of the Condition Monitoring Sensors
The installation of the condition monitoring sensors is vital for its performance. Incorrect mounting will most probably give you data that relates not only to a change in conditions but also to the instability of the sensor itself. Therefore making the sensor’s data unreliable.
Figure 8: schematic of frequency range and sensor mounting method
Several different mounting methods exist:
- Wax mounting: it is very convenient but we don't recommend using this method for mounting accelerometers. The inconsistency in thickness and the damping effect (low rigidity) of the wax make results unreliable at higher frequencies.
- Adhesive mount: very appropriate when stud mount is impractical or even not possible because drilling is not permitted.
- Magnet mount: appropriate for troubleshooting or periodical measurements. Magnetic mounting adapters are used to attach accelerometers to the ferromagnetic material.
- Stud mount: very appropriate for permanent and high-frequency vibration monitoring.
Vibration sensors should be mounted in locations that ensure the measurement of vertical, horizontal, and axial movement:
To detect imbalance and bearing issues, horizontal measurements need to be taken. In this case, the sensors should be mounted as close as possible to the motor bearings and pump bearings.
To detect looseness and problems with structural rigidity or foundation, vertical measurement needs to be taken with sensors placed close to the motor and pump drive end bearings.
To detect misalignment between the motor and the load, axial measurements need to be done. In this case, sensors should be attached close to the motor and pump drive end bearings.
Accelerometers should be mounted as close as possible to the source of vibration is measured. Mounting on a clean, smooth, flat, unscratched surface via a drilled and tapped hole is highly recommended to obtain a stable position of the sensor especially when measuring high-frequency vibrations. Make sure that the screw is not longer than the threaded hole. There should be no spacing between the sensor and the measured object.
In case drilling is not allowed on the machinery glue with metallic properties can be used. This ensures good vibration transition.
STEP 6: Data Collection and Data Interpretation
Machines talk to us but unfortunately, they don't speak English or any other language that humans understand. They communicate through vibration signals generated while the machine is running. Therefore it is essential to understand vibration to be able to assess the machine condition. But how do we do that?
To translate vibration signals into human-readable language we use the so-called vibration diagnostic tools composed of three main parts:
- Data Acquisition Hardware
- Condition Monitoring Software
Sensors are devices connected to the measurement point whose purpose is to detect physical phenomena events or changes and translate them to proportional electrical values. Different sensors are used in condition monitoring and range from displacement transducers, accelerometers, shock-pulse transducers, and velocity transducers.
Each of them fit for the purpose they have been developed for. The main difference is the accuracy they offer in a specific frequency range:
- Displacement transducers are very good in the frequency range from 0Hz to 200 Hz.
- The velocity transducers are a perfect fit for mid-range frequencies from 2Hz up to 1kHz.
- Accelerometers are the best in the range from 5Hz – 20kHz.
On top of the accuracy and frequency range there are also several other factors to consider when selecting the transducer like:
- Temperature range
- Dynamic range
- Mounting possibilities
- IP – Ingress protection
- Wireless or wired
To choose the right one, you need to understand what you are going to measure.
For measuring the displacement of stationary signals (DC) or very low-frequency signals displacement sensors are used called eddy current proximity probes. These sensors detect the inhomogeneities of metal material by sensing changes in a magnetic field generated by a reference coil. Proximity probes are used for non-contact displacement measurements and usually require to be permanently mounted on the machine.
Figure 9: Eddy Current transducer principle of operation
The most commonly used sensors to measure vibration are Accelerometers. There are several types of accelerometers. The most frequently used are:
- capacitive MEMS
- FBA (Force Balance Accelerometers)
- Piezoelectric Accelerometers (IEPE accelerometers)
Accelerometers can be Wireless or Wired. Wireless accelerometers are very easy to mount, since no cables are required, but are not at all appropriate where dynamic (high speed), real-time measurements are needed due to their battery life limitations.
Figure 10: Permanent single-channel Data Acquisition Module with a Vibration and Temperature sensor probe
Data Collectors – Data Acquisition Hardware
Data acquisition hardware's purpose is to convert electrical signals (analog) into digital.
Figure 11: Multiple-channel High-End Data Acquisition system
Key elements of a data acquisition system are:
- signal conditioning
- analog to digital converters (ADC)
- computer interface/bus
Signal conditioning is the part of the data acquisition unit circuitry that prepares the analog signal coming from the sensor to be ready to be acquired by the ADC. The signal conditioning circuitry transforms the signal by amplifying, filtering, attenuating, and possibly isolating it.
Analog to Digital converters (ADC)
ADCs are integrated circuits that transform the analog signal from the signal conditioning circuit into digital before they are sent to the computer for further computation. The main characteristics of an ADC converter are resolution and sampling rate.
It is a communication interface between the data acquisition unit and the computer. There are several options available on the market including PCI, USB, Ethernet, Wi-Fi on top of which different protocols can be implemented like TCP/IP, EtherCAT, ProfiNet, etc. The selection of the communication interface depends mainly on the required data rate, the spatial distribution of devices, and the environment (laboratory or industrial).
Similar to accelerometers, there are also a lot of data acquisition systems, composed of data acquisition hardware and software, on the market. Again there are big differences between them especially when we get deeper into the understanding of their working principle, reliability, software features, repeatability of measurements, method of use (portable or permanent/online), distributed vs. centralized, etc.
Condition Monitoring Software
Condition monitoring software can be purposely built for condition monitoring of specific machines or can be reconfigurable and thus appropriate for complex machinery diagnostics applications.
Figure 12: Dewesoft condition-based monitoring Paper Mill user interface
The software can be very basic and easy to use, offering just some overall statistical values. In other cases, the software can have all the necessary features to analyze the raw data obtained from the data acquisition devices. It can also have long-term historical data storage and trend visualization possibilities enabling users to detect all possible machine failures. The best software solutions offer also direct connectivity and data transfer to 3rd party distributed control systems over different available protocols.
There are differences between condition monitoring software providers also in terms of access to the data. Can be computer-based with local access or web-based software for remote monitoring.
Software tools for typical machine faults detection:
|Software Features||Typical Machine Faults|
|Shaft Unbalance||Misalignment||Loose foot/foundations||Stator eccentricity||Rolling Bearing defects||Journal Bearing wear out||Gear tooth failure||Cavitation||Turbine blades wear||Structural resonances|
|Frequency Response Function||✖||✖||✖||✖||✖||✖||✖||✖||✖||✓|
|Shaft centerline plot||✓||✓||✖||✖||✖||✓||✖||✖||✖||✖|
|Displacement, Velocity, Acceleration||✓||✓||✓||✓||✓||✓||✓||✓||✓||✓|
|RMS, Peak-Peak, Peak||✓||✓||✓||✓||✓||✓||✓||✓||✓||✓|
|Analysis Bandwidth and Resolution||✓||✓||✓||✓||✓||✓||✓||✓||✓||✓|
|Tachometer & trigger||✓||✓||✓||✓||✓||✓||✓||✓||✓||✓|
|Time Data Recorder||✓||✓||✓||✓||✓||✓||✓||✓||✓||✓|
|Time Data Editing||✓||✓||✓||✓||✓||✓||✓||✓||✓||✓|
|Operating Deflection Shapes||✖||✖||✖||✖||✖||✖||✖||✖||✖||✓|
STEP 7: Maintenance Tasks Determination
When you have all the data on your desk, to take appropriate maintenance actions, you need to interpret the acceleration, displacement, temperature and other data collected using the above-described software tools.
There are two ways of doing it:
- Manually with a qualified Predictive Maintenance Engineer (PdM). They can be employed or can be outsourced from companies offering PdM services.
- With automatic data interpretation using Predictive maintenance software solutions available on the market.
Both options have their benefits and drawbacks. Qualified PdM’s are normally expensive and spend a lot of time analyzing the data. However, once they know the machine in detail they can very reliably predict their failure and set appropriate scheduled maintenance tasks well in advance.
On the other hand, predictive maintenance software solutions are way less expensive but by far less reliable. Just think how many different types of equipment exist, in how many different environments and conditions they work. Because of that, it is impossible to set a unique baseline for all of them and set unique thresholds for alarming.
Today the best solution probably is a combination of automatic, used for non-critical, and manual for critical machinery.
The efficiencies of mechanical equipment can be increased by using proper Machine Condition Monitoring solutions.
The choice of the system depends on the criticality of the asset, cost of replacement/failure, asset access possibilities, cost of monitoring and expected fault progression rate.
Portable, low-cost systems are normally used for non-critical assets with low-replacement costs and slow fault progression rates.
However, big advances in technology lately have enabled permanent monitoring solutions to become very cost-effective. Due to the cost decrease, higher reliability, and work efficiency more and more customers opt for permanent solutions.