In 1901, a patent was issued to Ransom E. Olds for the idea of a continuously moving assembly line, which he used to build the first Oldsmobile vehicles. In 1913, Henry Ford improved the concept by adding moving conveyor belts and with these two innovations, the time needed to assemble a car went from 1½ days to 1½ hours. The modern assembly factory was born.
During the next four decades, the idea of a moving production line was adopted by numerous industries, from radios to razors, clocks to cribs, nails to newspapers. During World War II, the USA built 300,000 aircraft using moving production lines. The idea became the backbone of manufacturing around the world and remains the primary way to provide products to mass markets.
The tools and machinery used to do assembly work are the most critical part of any line. In the 1950s, the reliability of factory equipment was becoming an important part of operations. If a single machine stopped working, the entire line would shut down until repairs could be made. Idle machinery and assemblers reduced efficiency and impacted costs. Maintenance became important.
Throughout the last half of the 20th century, most factory maintenance consisted of scheduled periodic maintenance. This improved the reliability and service life of the machines but did nothing to help factory operators predict and avoid any equipment malfunctions or unscheduled downtime. Over the last twenty years, machine condition monitoring has become an important part of factory operations. Sometimes referred to as Smart Factory, Industrial IoT (IIoT), or Industry 4.0, factories are adding sensors and analytical systems that keep tabs on the health of the manufacturing equipment.
Condition monitoring is the process of outfitting equipment with sensors that can identify significant parametric changes indicative of an imminent fault or failure. These systems look for two things. First: Is the machine operating outside of its design parameters, requiring a quick response? Second: What are the long-term trends of critical parameters for predicting when the machine will need maintenance, repair, or replacement.
Factory Electric Motors
One of the most ubiquitous machines in factories is the electric motor. It’s estimated there are 300 million motors running in manufacturing facilities around the globe today. They provide the power to fabricate, assemble, and move products in a factory. The failure of a single motor can bring a production line to a grinding halt, costing $10,000s per hour. Condition monitoring sensors and analytics give an early warning of problems. Treating these problems quickly results in a lower cost of maintenance, fewer disruptions in process flow, and improved safety for the equipment operators.
The health of any motor can be determined by monitoring three operating parameters:
Vibration — measured with an accelerometer.
Temperature — measured with a contact type temperature sensor.
Current — measured with an inductive or shunt type current sensor.
Accelerometers provide data regarding the motor’s mechanical health. Temperature sensors provide data regarding both mechanical and electrical health, and current sensors look at the electrical health. Sensors can also provide data regarding the condition of equipment and tools attached to the motor.
All rotating motors have an armature that’s suspended by bearings and rotates at various speeds. The armatures are balanced so as not to vibrate while rotating. However, an unbalanced or damaged armature creates vibration that can affect bearings. If the vibrations are at resonant frequencies of other equipment attached to the motor, damage can be severe and quick. Vibration frequency from a motor coincides with the motor RPM and typically ranges from a few Hz to 4 kHz. Accelerometers are suited for measuring these vibrations.
Accelerometers are also used as contact microphones that listen for high frequency noise (squeal) from bearings where lubrication has been depleted. These frequencies range from 5kHz to 15kHz. Piezoelectric accelerometers work well because they have a broad frequency response that covers both frequency ranges with a single device.
Typically, a Fast Fourier Transform (FFT) is used to analyze the signal. The FFT data shows each frequency band of vibration and its intensity. Figure 2 indicates the natural vibration frequencies and amplitudes of a typical motor.
New motors have natural vibration frequencies that are recorded and used as a baseline to compare to data taken later in the motor’s life. If these measurements stay in the same range as the original data, the motor is in good health. If frequencies or intensities begin to shift with time, it’s an indication of wear in parts of the motor and a failure may be imminent. Repairs can therefore be scheduled before a failure occurs. This eliminates catastrophic line shut-downs and improves manufacturing efficiency.
Figure 3 shows the model 8911 vibration sensor from TE Connectivity Sensors Business Unit. It contains a piezoelectric accelerometer and a microcontroller that performs an FFT conversion to a format that customers need. In addition, this sensor is battery powered and transmits the data wirelessly over LoRa frequencies. This eliminates hard wiring. The sensor can be easily mounted to a motor with an adhesive, mounting stud, or a magnetic base. With no wiring and simple mounting, the sensor can go from out-of-the-box to full operation in a very short period of time and at very low cost.
The two parts of a motor most affected by high temperatures are bearings and stator windings. To be effective, sensors that measure these parameters must be in intimate physical contact with the motor parts of interest, so unique designs are usually required for motor applications.
In operation, motor bearings typically run between 60° – 70°C (140° – 160°F). When properly loaded and lubricated, bearings can have extremely long service life. The most common problem occurs when lubricants disappear because of leaks or fail due to ambient overheating. These failure modes can occur over fairly long periods of operation — 1000s of hours. As lubricant is depleted, the bearing temperature begins to rise and can be easy to track with a temperature sensor and analytic software.
Figure 4 shows a spring-loaded NTC or RTD type temperature sensor that’s designed to be in direct contact with an outer bearing race. This mounting provides the most accurate temperature measurements. The bearing housing must have features designed in that will accommodate the temperature sensor and provide close access to the bearing.
Temperature sensors are also used to measure the temperature of stator electrical windings. These are the coils that generate the magnetic fields to propel the rotation of the armature. If these windings are damaged by impact or corrosion, the resistance of the wire increases, and their ability to conduct current diminishes. This causes the affected wires to heat. They can get to temperatures that will melt insulation and even cause a fire.
Figures 5a and 5b show temperature sensors (TE Connectivity Sensors Business Unit) being built into stator windings on a motor. The sensors become a permanent part of the device.
As with the vibration sensors, initial operating temperature is measured and recorded. As time passes, the temperatures are taken periodically and compared to the baseline. If the temperatures begin to deviate from normal, it’s an indication of mechanical problems for the bearings or electrical problems for the stator windings. Again, maintenance can be scheduled before a major failure occurs, preventing an unexpected line shutdown.
While vibration and temperature sensors can report the physical health of a motor, a current sensor can check its electrical health. Current sensors come in two basic configurations: shunt or inductive designs. Each has advantages and disadvantages. The amount of current a motor draws is affected by many things: the applied voltage, the motor speed, the load on the motor shaft, and the condition of the armature brushes. Out-of-range currents can signal power voltage problems, shorts in the stator or armature windings, worn brushes, or problems with attached tooling.
A shunt current sensor is a very low value power resistor placed in series with the electric power line to the motor. The voltage developed across the resistor is proportional to the current flowing through it. Using Ohm’s law (I = E/R) and knowing the voltage and resistance values, the current can be calculated. Shunts are used where very high accuracy is required or for high frequency brushless motors. Figure 6 shows the schematic for a shunt sensor.
Inductive current sensors use a high permittivity ring around the power wire. Current through the wire induces a magnetic field in the ring (Bin). A Hall effect sensor embedded in the ring measures the magnetic field, and through an amplifier and coil wrapped around the ring, induces a reverse magnetic field (BF) to counteract Bin. The voltage required to create the BF field is the output signal. The net effect of the two magnetic fields will appear as a null field at the Hall sensor. The higher the current flow in the power wire, the higher the BF voltage must be. (See Figure 7)
Beyond the Motor
Often, equipment attached to the motor can also be monitored for health. Figure 8 shows an example of a milling machine that uses a multi-fluted rotary cutter to machine a flat surface into a piece of work. The condition of the rotary cutting tool is very important to the quality of the cutting operation. The sharper the flutes on the cutter, the more precise the cut, and the better the surface on the finished work. As each flute makes a cut, the motor current increases to provide the needed power. Sharp flutes require less power to cut the metal. Dull flutes require more power.
By monitoring the motor current for this milling machine, it’s possible to determine which of the flutes on the cutting tool are sharp and which are dull and need to be sharpened.
The Contributions of Condition Monitoring to the IIoT
From the invention of the production line and the industrial revolution it created, there have been numerous technical innovations that have pushed progress in manufacturing. Factories have experienced average productivity gains of 2.8% per year over the last 70 years. Technologies such as power tools, injection molding, pick & place equipment, automatic testing, and robotics have all contributed. Today, machine condition monitoring is being added to the mix. Successful condition monitoring starts with sensors located on factory machines constantly reporting data on their health and operation. By using real-time analytics, factory operators can identify problem spots and schedule the needed repairs and maintenance without waiting for catastrophic failures. This will provide productivity gains to manufacturing. It all starts with sensors, however.