Calibration of Conventional Systems and Wireless Sensors
In the concept of Industry 4.0 is the digitalization of processes, and in this context is the sensoring of industrial machines.
Sensoring allows the monitoring of assets and can be used by predictive maintenance in decision making. From parameters collected from machines, such as vibration and temperature, it is possible to monitor the condition of an asset in real time.
Through wireless sensors, places that previously, for reasons of access or security, were not susceptible to data collection, are now points that can be easily monitored.
And this is precisely Dynamox’s core business, developing wireless monitoring systems for industrial assets.
As the use of such sensors expands, some paradigm shifts are inevitable as the elimination of the need to travel to the machines, by field inspectors, to collect parameters.
With the use of wireless sensors installed in these machines, this process can be automated and inspectors can be allocated to other tasks, such as analyzing the data generated.
Significant increase in the number of devices
In the traditional data collection method through offline collectors used by inspectors on their routes, each company has one or a few units of these devices.
In wireless monitoring, with sensors that are fixed to the machines, it is common for industries to have hundreds or thousands of this type of sensor, since each one is responsible for monitoring a specific point of an asset.
Changes in the calibration process
The calibration process of signal acquisition systems is a vital part for the assertiveness of the data collected, because it ensures the accuracy of the measurement system as a whole.
This calibration process is facilitated in traditional manual monitoring systems that we have previously commented on, because there are a few devices to be calibrated, while in automated monitoring systems, with thousands of sensors fixedly installed in industrial machines, this process becomes unfeasible.
In the case of offline collectors in the traditional method, calibration must take into account the sensor (usually piezoelectric and analog), the cables, connections (if any) and the collector, since calibration of the sensor alone guarantees more reliable collection data, but does not guarantee non-contamination of noise in the collections and, consequently, the measurement as a whole.
On the other hand, calibration of wireless automated monitoring systems only considers the individual sensor and therefore has less uncertainties associated with the measurements.
In this text, we will give more details on how the calibration process works for the DynaLoggers (wireless sensors for collecting vibration and temperature developed by Dynamox).
But first, let’s better understand how the calibration process of traditional systems (offline collectors) works.
Calibration of Conventional Systems (offline collectors)
The verification process for conventional systems is focused on the calibration of the sensor (accelerometer), which is regulated by ISO 16063, with emphasis on the following parts:
- 16063 – Part 11. Primary vibration calibration by laser interferometry
- 16063 – Part 13. Primary shock calibration by laser interferometry
- 16063 – Part 21. Vibration calibration by comparison method
- 16063 – Part 22. Shock calibration by comparison method
These norms standardize calibration based on obtaining the sensitivity and frequency response of the sensor.
In technical terminology, an accelerometer is a transducer, which means that it transduces mechanical energy (vibration) into electrical energy (electrical charge), where sensitivity can be understood as the ratio of this transduction.
Thus, this approach assumes that the sensor has analog output by nature, as shown in Figure 1.
The reference accelerometer (calibrated and highly accurate) and the uncalibrated accelerometer are exposed to a predefined vibration signal generated by the shaker.
By using the analog-to-digital converter the electrical voltage signal generated by both accelerometers is captured.
The sensitivity (Ss) relates the voltage (V) generated by the sensor to the acceleration value (Ar) to which it was exposed (as measured by the reference accelerometer), thus obtaining a sensitivity with the equation below:
When drawing a parallel between the calibration of traditional systems and the application context of Dynaloggers, two issues must be considered:
- Since the sensors are permanently installed in the machines they monitor, it would be necessary to remove them from where they are installed. This means stopping the operation of the asset to remove the device(s) and then applying the calibration procedure. This point becomes even more unfeasible if we consider the number of sensors that several of our customers have.
- The calibration method proposed by the norm cannot be applied to the accelerometers used in the Dynaloggers, because they have sensors with digital output (see Figure 1). Unlike traditional accelerometers, whose sensitivity is calculated in units of volt per gravity (V/g) or volt per second squared (V/m/s²), the accelerometers used in the DynaLoggers have less significant bit units per unit of acceleration (LSB/g or LSB/m/s²).
Currently there is no standardized process for calibrating acceleration sensors with digital output. To better understand the particularities of this type of sensor, we will detail the operation of these digital accelerometers below.
Digital MEMS Accelerometers
The DynaLoggers are equipped with MEMS (Micro-Electro Mechanical Systems) accelerometers made of silicon with micromachining technology and are already consolidated models in the market, marketed since mid-2015.
MEMS sensors have mechanical structures, which can be described as a spring-mass system, where the spring deforms generating a capacitance variation that, on its part, generates an electrical voltage signal proportional to the acceleration.
The most significant difference between the conventional sensors is that the signal conditioner and analog-to-digital converter are integrated into the accelerometer circuit.
In this signal conditioning, filters and processing are applied to ensure the accelerometer’s frequency response.
In addition to the miniaturization of the vibration sensor, MEMS technology has great reliability in relation to drops and impacts on the sensor.
Some studies show that the most common failure mechanism in MEMS accelerometers is caused by fatigue, which means that device degradation occurs most commonly due to sensor usage issues.
Researchs  also show that the mean-time-to-failure (MTTF) due to fatigue effects of the mechanical elements of a MEMS accelerometer is about 1.9×1081 s (worst case) , which indicates that MEMS sensors have a high resistance to mechanical fatigue.
Even under accelerated degradation conditions (at high temperatures, frequency, and amplitude), the estimated failure rate is only a little compromised .
Therefore, based on these robustness studies it is expected that with an estimated lifetime between 3 and 5 years, a DynaLogger will not change its response within this period.
How does the calibration of the DynaLoggers work then?
All sensors present in the DynaLoggers, whether accelerometer or temperature sensor, are calibrated by their respective manufacturers.
The manufacturers of the MEMS accelerometers used in the DynaLoggers correct the gain and sensitivity of each accelerometer with compensation routines stored in the internal memory of the sensor.
Each time the device is triggered, the compensation values are loaded, obviating the need for subsequent calibrations.
In addition, the sensors undergo verifications within our strict quality process, where Dynamox is ISO 9001 certified. The sensors are tested in two steps during the production process.
The first test procedure is concerned with checking the frequency response of the accelerometer and the assertiveness of the digital sensitivity.
In this procedure, also called back-to-back, we use a more sensitive and calibrated accelerometer together with a device that generates vibration signals to expose the DynaLogger to a known signal.
By comparison, the values measured by the DynaLogger and the values measured by the calibrated accelerometer are evaluated to determine the collection’s errors and uncertainties.
This process is done on a sample basis before the production of each batch of DynaLoggers.
The second procedure is to ensure that the DynaLogger assembly process does not interfere with the accelerometer’s response.
Thus, using a specific bench, several DynaLoggers are exposed to a known vibration signal that can excite up to 20 DynaLoggers and their three axes simultaneously.
This process in batches is done daily during the production process.
Besides these calibration steps, the quality testing still continues.
The DynaLoggers are installed to monitor an asset and the system that manages the temperature and vibration sensors within the DynaLoggers performs verifications on the measured data.
Whenever the system detects that inconsistent data have been acquired, it discards the data, preventing false values from being presented to the analysts.
DynaLoggers’ responde after a long period in use
For the purpose of checking the working conditions of the DynaLoggers in operation, we used as an example two DynaLoggers that were installed on one of our customer’s machines.
During the production of the devices, that is, before installation in the field, a sample of the batch was measured for quality inspection, and the results were saved in Dynamox’s database.
With these DynaLoggers, shown in the photos below, back-to-back comparisons were performed with a calibrated reference accelerometer, and the frequency response function of the DynaLoggers was estimated.
The figure below shows the comparison of the response of these DynaLoggers before and after installation and operation in the field.
As it can be seen in the graph, despite the fact that the DynaLogger was clearly installed in a very aggressive environment, with the presence of cracks being observed, marks of contact with chemical products and slight oxidation, it can be seen that the frequency response of this DynaLogger did not change significantly, therefore evidencing the robustness and assertiveness of the calibration and testing method adopted.