Vibration Monitoring using Arduino Data Acquisition to Detect Anomalies in the Operation of Water Pump.

Arduino Data Acquisition (ADAQ) is a popular development device among hobbyists nowadays because of its cheap price. Arduino comes in many forms, and recently, Arduino has been used in many industries for many applications, including vibration monitoring. This study aimed to investigate feasibility of using an Arduino data acquisition to detect anomalies in the operation of a water pump. An Arduino Mega was used with an ADXL345 accelerometer attached to a water pump motor to measure vibration continuously. Vibration was measured with a sampling rate of 530 Hz. Water pump vibration data abnormal conditions were measured and analysed to determine the problem. Results showed that vibration during anomalies was significantly greater than that in normal conditions. This study concludes that Arduino Data Acquisition with ADXL345 can detect abnormal water pump running conditions.


Introduction
The cold water system in buildings is a basic and important system that is compulsory to install in every building.The necessity for a consistent water supply from these systems is critical, given that individuals rely on water for a multitude of everyday activities, including drinking, food preparation, cleaning, toilet use, and various domestic purposes.In the case of high-rise buildings, water distribution often relies on pumping systems due to the structure's height or inadequate water pressure from the source.In such scenarios, ensuring the continuous upkeep of the pump system is imperative to avoid disruptions in the water supply for the building's occupants.The malfunction of a water pump system not only causes inconvenience for users but also has a substantial impact on maintenance expenses.Therefore, it is crucial to regularly maintain the system to extend its longevity and preserve its effectiveness.Nevertheless, routine maintenance is insufficient, as pump faults or anomalies during operation can arise unexpectedly without immediate detection.
The breakdown of a water pump system may be caused by various factors, including wear and tear, aging components, bearing failure, shaft misalignment, and excessive physical stress [1,2].While these factors are detectable through physical examination, some water pump issues, such as cavitation or water hammer, are not immediately visible due to their specific characteristics.Cavitation in a water pump is triggered by low pressure, leading to the formation of damaging bubbles when they collapse.Similarly, water hammer, a hydraulic phenomenon, results in sudden pressure surges due to valve closures or fluid flow changes.These events can potentially damage the pump and the entire system, underscoring the need for implementing safety measures and regular maintenance.Most of the problems stated are difficult to detect physically but possible if we monitor vibration signals [1].Vibration signals serve as indicators of a machine's mechanical condition, with each mechanical problem or defect producing unique vibrations.Therefore, assessing the vibration signal can help pinpoint the underlying cause and facilitate the necessary repairs.Understanding the current state of the water pump and predicting potential failure points is crucial for risk mitigation.By measuring the water pump vibration and analysing the data in both time and frequency domains, the condition of the water pump and its associated components, can be evaluated.Analysing the vibration data enables the determination of the current condition of the water pump and aids in anticipating future maintenance needs.Moreover, incorporating vibration monitoring techniques can help detect early signs of cavitation and water hammer, enabling timely intervention and preventing severe damage to the pump and associated components.Such proactive measures are essential for ensuring the smooth and efficient operation of water pump systems.
Vibration data from the water pump motor serves as a valuable resource for maintenance, particularly in the implementation of preventive maintenance strategies.Preventive maintenance encompasses regular maintenance scheduling and continuous equipment monitoring to prevent unforeseen breakdowns.The primary objective of scheduled maintenance is to restore the equipment to its original state, ensuring optimal functionality and an extended operational lifespan [3,4,5].Utilizing the condition-based maintenance technique relies on data acquired from condition monitoring to offer maintenance recommendations [6].Maintenance activities are undertaken upon reaching predetermined thresholds, guided by specific indicators.Neglecting the current machine condition within the context of preventive maintenance increases the likelihood of unexpected and inevitable machine failures, leading to unnecessary expenses [7].
The importance of predictive maintenance for industrial equipment has intensified in the Industry 4.0 era, leading to the adoption of IoT platforms as a practical solution.A real-time monitoring system for electric motors was developed using affordable hardware and open-source software, focusing on vibration analysis in the temporal and frequency domains to identify operational irregularities and enable predictive maintenance [8].Similarly, an IoT-based system was established, incorporating sensors like accelerometers, current sensors, and thermocouples to monitor parameters such as temperature, vibration, and current, thereby assessing the health of electric motors [9].Additionally, an IoT-based platform was designed for continuous monitoring of induction motor parameters, providing alerts and data access for predictive maintenance purposes [10].Another study explored the use of machine learning algorithms in conjunction with traditional techniques like modal analysis and wave propagation to detect looseness in bolted joints.The review examines the advantages, limitations, and applications of these methods, highlighting the potential of IoT-based health monitoring for remote monitoring of bolted connections [11].
At the moment, vibration analysers are the most commonly used condition monitoring tools.Depending on the specifications, the vibration analyser may be rather costly.To predict the condition of a machine, this device has to be installed in all machines under a person's supervision, and vibration needs to be measured at all times every day.Many analysts, therefore, have looked into a different alternative for this purpose.One of the most popular platforms for electronic development is an Arduino microcontroller.Arduino is a platform with open access and affordable hardware [12].It features a large number of sensors that are specifically made for different uses, and they are reliable and durable even in harsh environments.A Wi-Fi module can be used with Arduino to link it to the cloud, enabling online monitoring for specific circumstances.The Arduino platform can also function as a data-collecting device for vibration monitoring.
This study will assess the performance of an Arduino data acquisition to monitor water pump conditions to detect anomalies, especially due to cavitation or water hammer issues.The study developed a vibration measurement device using an Arduino microcontroller and an ADXL accelerometer that was installed at a water pump motor for monitoring purposes.The device was connected to WiFi for online monitoring, which sent vibration data online and could be used as an online monitor to detect anomalies such as water hammer and cavitation problems.

Methodology
The experimental setup was using an Arduino microcontroller and an ADXL345 accelerometer which in this paper be called as the Arduino Data Acquisition system (ADAQ).This project was carried out in the following steps in order to accomplish the goals stated in the preceding section: x Design and Development of ADAQ x Verification using a Vibration Calibrator x ADAQ Installation on Water pumps Motor x Data analysis

Design and Development of ADAQ
ADAQ, as depicted in Figure 1, was set up and used for this study.It consisted of an Arduino microcontroller (Arduino MEGA), a microSD card adapter, a clock module, a base shield, and an ADXL345 Digital Accelerometer.The ATmega2560 microcontroller that powers the board was linked to an ESP8266 Wi-Fi module to provide Wi-Fi connectivity.To get started, the board has the option to be powered by an AC-to-DC adapter or battery or linked to a computer using a USB cable.For the purpose of this study, raw data was collected which stored in a microSD card soldered on a shield equipped with a Wi-Fi ESP8266 module.For easy connection between sensors and Arduino LCD 16x2 To display selected output The ADXL345 accelerometer could sample at a maximum rate of 530 points per second which was limited by the components, instructions and ability of the microcontroller.The data was only stored for 300 points, or less than one second, due to the dynamic memory's limitations.

Verification using a Vibration Calibrator
The ADAQ accuracy to measure vibration was verified by comparing the data measured with a vibration generated by a calibrator.In this study, it was verified with a B&K Accelerometer Calibrator Type 4294 at 159 Hz with a magnitude of 10 m/s 2 .Once verified, the ADAQ was used for the purpose of this study.ADAQ was also previously verified with a piezoelectric accelerometer from previous study [13].

ADAQ Installation on water pump Motor
The ADXL345 accelerometer was installed on the water pump for vibration measurement in the water pump room, as illustrated in Figure 2. The water pump is a vertical multistage centrifugal type with a power rating of 3.0 kW and operates at a speed of 2900 rpm.Vibration measured by the ADAQ was converted to a r.m.s.value before being sent to the cloud system.The raw vibration was stored on the SD card, as previously explained.The measurement was triggered if the pump started based on a threshold set in the ADAQ.The measurement was repeated so long as the pump was in operation.Due to the limitations of the ADAQ, each measurement was less than 1 second and required about 30 seconds to complete all the instructions before the next measurement.

Data analysis
The vibration measured was converted to root-mean-square (RMS) acceleration before sent to the cloud.The data also was stored in the SD Card which was later retrieved for time domain and frequency domain analysis.On the 3 rd April 2023, it can be seen that the vibration RMS value in x, y, and z values surged up to a higher than normal value on the previous running operation, starting at about 11.37 am, and continued to vibrate abnormally on the 4 th April 2023 with the vibration maximum RMS value reaching up to 2.89 m/s, as shown in Figure 5.

Time domain and frequency domain vibration of the water pump
Further analysis employing both time domain and frequency domain techniques for the water pump revealed that under normal conditions, the pump operated at a frequency of approximately 48.5 Hz, corresponding to a rotational speed of around 2900 RPM.Notably, significant vibrations were observed at the 1x, 3x, and 5x frequencies of the water pump rotation, exhibiting amplitudes of 0.02 m/s², 0.002 m/s², and 0.005 m/s², respectively, as illustrated in Figure 6.The data depicted in Figure 7 indicates irregular pump operation during the anomalies, as revealed by the time domain analysis.Conversely, the Fast Fourier Transform (FFT) spectrum for the healthy pump as observed in Figure 6 exhibiting minimal peaks signifying its normal operation, whilst the FFT observed in the unhealthy operation of the pump showed the significantly high signals as shown in Figure 7.The pump displays high-amplitude and random-frequency vibrations at higher frequencies, as depicted in Figure 5, distinguishing it from the behaviour of the normal pump.Sudden changes in acceleration value in time domain analysis could indicate cavitation effects.These could indicate that cavitation happened in the water pump during the detected anomalies [14].The findings of this investigation make a significant contribution to the field of water pump monitoring, particularly with regard to the identification of cavitation.The ADXL345 accelerometer combined with Arduino Data Acquisition (ADAQ) has shown to be a practical and affordable solution for identifying irregularities in the water pumps during operations.When compared to normal conditions, the mean amplitude (RMS) increases during cavitation occurrences, which is a strong indication of anomalous pump activity.This finding is significant because it offers an accurate and measurable criterion for identifying cavitation, a condition that is known to harm pump components.
In addition, the conversion of time domain data to fast Fourier transform (FFT) further strengthens the argument for the capability of this device in detecting cavitation.The exposure of high- amplitude and random-frequency vibrations at higher frequencies during cavitation events emphasizes the ability and sensitivity of the Arduino-based system in capturing subtle changes in pump performance.The practical impact of the results of this study includes various industries that depend on water pump systems.Timely detection of cavitation is important for preventive maintenance, minimizing downtime, and preventing long-term damage to the pump.The capabilities and accessibility of the Arduino platform make this solution very attractive for widespread implementation in scenarios where cost considerations are paramount.
Overall, the results of this study not only confirm the realization of the use of Arduino Data Acquisition with the ADXL345 accelerometer for cavitation detection, but also open up opportunities for cost-effective and efficient water pump system monitoring in various applications.This research forms the basis for the adoption of readily available technology in an industrial context, encouraging a proactive approach to pump maintenance and system integrity.

Conclusion
This study utilized ADAQ to measure vibration conditions within a water pump's motor, with findings that were determined to be both realistic and indicative of the actual condition of the water pump.The measured data was subsequently extracted and analysed using the Fast Fourier Transform (FFT) method.The results provided insights into the water pump's operating conditions over a 3-day period, prior to conducting further comprehensive analyses in both the time and frequency domains.The study underscored the efficacy of ADAQ in capturing vibration data during both the normal functioning and anomalous operations of the water pump.The data derived from ADAQ proved instrumental in facilitating FFT analysis, thus enabling a comprehensive evaluation of the pump's condition.Moreover, it was established that with more extensive analysis, detailed identification of the underlying issues within the pump could be accomplished.

Figure 2 .
Figure 2.An accelerometer was installed at the water pump motor.

Figure 4 .
Figure 4. Vibration measured at the water pump on 3 rd April 2023.

Figure 5 .
Figure 5. Vibration measured at the water pump on 4 th April 2023.

Table 1 .
Arduino Data Acquisition System Components.