Design of a Load Monitoring System for a Distribution Transformer

In this study, a load monitoring system was developed to monitor the loading conditions of a distribution transformer over a week. Testing revealed that the transformer’s peak load reached 14.96 kVA, pushing it to operate at 149.55% of its 10kVA rated capacity, indicating an overloaded condition. The data visualization of the load monitoring system also indicates that the recorded daily peak loads consistently exceeded the transformer’s capacity, suggesting the necessity of uprating the transformer. Utilizing a split-core current sensor, the load monitoring system is implemented without the need to de-energize the distribution transformer from operation. It effectively assesses the distribution transformer loading conditions in real-time and remotely, allowing for continuous and efficient data collection without on-site manpower requirements.

Introduction

Distribution transformer is one of the most important equipment in power distribution system network. The data assets and provision of transformers are very important aspects of the electric network, as a huge number of distribution transformers are distributed over a huge area [1]. The average lifespan of a transformer is about 25 to 35 years, but they fail in large numbers due to various reasons, causing huge economic and service impacts on power consumers [2]. Failures of transformers not only impact industries and consumers but also the economy of the country affected by the same, causing social and political ramifications. Some of the most common causes of distribution transformer failure are: (1) prolonged overloading, (2) unbalanced loading, (3) poor maintenance, and lack of monitoring of transformers [3].

In this paper, the researcher will focus on the development of a monitoring system that can identify the real-time loading of a distribution transformer, giving information on whether the distribution transformer is operating at normal loading conditions, underloaded, or overloaded. The loading of a distribution transformer can be determined by dividing the measured load to the transformer’s nameplate rating. If the transformer is overloaded in a period of time, this could break the transformer’s insulation or damage the coils inside the transformer [4]. With the monitoring system in place, the collected information will help in making judgements whether the concerned distribution transformer requires uprating or downrating to prevent failures.

Sajidur et al. [5] developed a system to monitor the health status of distribution transformers in real-time, addressing the challenge of manually monitoring a large number of transformers distributed across wide areas. The system utilizes a mobile embedded system integrated with sensors and a GSM modem to monitor load currents, overvoltage, transformer oil level, and temperature. When abnormalities are detected, the system sends SMS messages to designated mobile phones, enabling utilities to optimize transformer usage and identify problems proactively. Since this system uses SMS only for alarms and notifications, it is not possible to observe the actual loading conditions of the distribution transformer. This missing feature can be addressed by implementing a database and a data visualization tool.

Romalyn et al. [6] assessed the electrical percent loading of distribution transformers used in a barangay. The researchers collected the energy consumption in kWh of connected consumers in each transformer and performed data analysis to determine the percent loading. Since the assessment is done using the collected monthly consumption, it is not able to determine the daily peak load of the distribution transformer. This concern is acknowledged by the researchers. Thus, the researchers recommended using a monitoring system on a daily basis.

Muruganandhan et al. [2] propose a system that focuses on autonomously monitoring the current, voltage, oil level, and winding temperature of a distribution transformer using sensors and PLC. The voltage and current sensors are used to monitor the secondary current and voltage of the distribution transformer, while the energy measured from the distribution transformer is sent to and monitored by the energy sensor. The authors used PLC and SCADA for monitoring purposes. In this paper, the author proposed a system using a Wi-Fi-enabled microcontroller for ease of maintenance and a database for recording the data.

Sahil et al. [1] designed a GSM-based monitoring system that provides SMS notifications once abnormalities such as overloading and a rise in temperature are detected in the distribution transformers. Although there is a delay of about 5–10 s between abnormality detection and sending notifications, the system is successful in helping to identify the failure in the distribution transformer. Since it is GSM based, retrieving data for data analysis and visualization is not possible. As such, the researchers recommended, for further development, to use a server module for receiving and storing data.

Tosin et al. [7] designed a monitoring system that continuously checks the temperature of transformers and predicts faults such as overheating or overcurrent. When abnormalities are detected, the system sends fault diagnoses to a base station via a GSM modem. Key components of this system include the ATmega328 microcontroller, GSM modem, LM35 temperature sensor, and ACS712 current sensor. Using ACS712 current sensor poses difficulty for implementing in an already existing distribution transformer since this sensor requires direct cable connection for current measurement. In this paper, a split-core current sensor will be used to address this issue and to allow measurements even if the distribution transformer is on an energized condition.

System Description

The block diagram and schematic diagram of the proposed Load Monitoring system are shown in Figs. 1 and 2, respectively. The monitoring device is installed in the secondary side of the distribution transformer, and the measured data is sent to Google Sheets via an internet connection using a portable Wi-Fi. The data is then linked to a data visualization tool to display the loading of the distribution transformer.

Fig. 1. Block diagram of the load monitoring system.

Fig. 2. Schematic diagram of the load monitoring system.

Hardware Component

Microcontroller

For this research, the microcontroller ESP32, as shown in Fig. 3 was used. It has a built-in Wi-Fi function, which makes it suitable for sending data remotely. It has the following specifications: two low-power Xtensa® 32-bit LX6 microprocessors, 448 Kbytes ROM, 520 Kbytes on-chip SRAM, 8 Kbytes SRAM in RTC SLOW, 8 Kbytes SRAM in RTC FAST, 1 Kbit of EFUSE, 256 bits MAC, Wi-Fi 802.11, Bluetooth v4.2.

Fig. 3. Microcontroller (ESP32).

Current Sensor

For this research, PZCT-2 current sensor is used to measure the current across the distribution transformer’s cable, as shown in Fig. 4. It is a current transformer which can measure up to 100A across cables with a diameter of up to 16 mm. The two output wires of the current sensor are connected to the AC communication module for processing the measured current. Three PZCT-2 current sensors are installed, which allow to measure current across a total of 3 cables within the same circuit group. The current measured by each current sensor will be processed separately by each module, and through the microprocessor, it will be sent to Google Sheets, where it will be analyzed. Fig. 8 shows sample google sheets data.

Fig. 4. Current sensor (PZCT-02).

Fig. 8. Sample google sheets data.

AC Communication Module

PZEM-004T AC communication module, as shown in Fig. 5 is used, which is capable of measuring and processing voltage, current, active power, power factor, frequency, and active energy. The AC communication module will be fed with voltage from the distribution transformer; as well as the current measured using the current sensor. It will then process these inputs and send the data to the microprocessor. The wiring diagram for PZEM-004T is shown on Fig. 6. For this research, a total of 3 modules are used to measure currents across 3 different cables, which is useful for distribution transformers that have multiple secondary cables.

Fig. 5. AC communication module (PZEM-004T).

Fig. 6. Wiring diagram for PZEM-004T.

Wi-Fi Modem

The Wi-Fi modem, as shown in Fig. 7, is inserted with a local SIM card to allow the microcontroller to send data to Google Sheets. It works with 3G and 4G networks with Wi-Fi transmission rate of up to 150 Mbps.

Fig. 7. Wi-Fi modem.

Power Supply

The load monitoring system is equipped with a 30,000 mAh, 12 V battery as power supply. The battery is then charged during the daytime using a 150 W solar panel. Using a Mini360 DC-DC buck converter, the DC voltage is converted to 5 V, which is the required voltage for the circuit.

Database

The load monitoring system uses Google Sheets as the database for the collected data. As the power is supplied to the microprocessor, it collects the data from the AC communication module and sends it to Google Sheets, where the sampling rate is set every minute. Data logged includes timestamps, sensor number, battery voltage, measured current, measured voltage, frequency, and power factor. The database is linked to Google Looker Studio for data visualization. Excel file version of Google Sheets can also be downloaded for further analysis.

Algorithm for the Load Monitoring System

1. Start
2. Turn on the system by pressing the power button on the solar controller.
3. Electrical parameters are measured and processed in the AC Communication module.
4. Values are passed to the ESP32 microcontroller.
5. The microcontroller, while connected to Wi-Fi, sends these values to Google Sheets.
6. Data recorded in Google Sheets is processed for visualization in Google Looker Studio.
7. End

Testing and Results

Fig. 9 shows the internal assembly of the load monitoring system enclosed in an IP65 junction box. It is then placed inside a customized enclosure to allow fixing to a utility pole. After assembling the load monitoring system, it was installed on a utility pole for testing, as shown in Fig. 10.

Fig. 9. Snapshot of the load monitoring system internal assembly.

Fig. 10. Load monitoring system installed on a utility pole.

The accuracy of the prototype’s measurement was checked prior to field testing. The measurement results of the prototype were compared to the measurement results using calibrated devices: digital multimeter for voltage measurement and clamp ammeter for current measurement. Fig. 11 shows the setup for accuracy testing. Table I shows the comparison of results between manual measurements using calibrated devices and the prototype.

Fig. 11. Setup for accuracy testing.

Table I. Manual vs. Protoype Measurement

No	Manual		Prototype
	Current	Voltage	Current	Voltage
1	1.94	238	2.08	238.1
2	1.94	238	2.08	238.3
3	1.94	238	2.08	238.9
4	1.94	238	2.08	238.0
5	1.93	238	2.07	237.5
6	1.95	239	2.09	239.3
7	1.96	240	2.08	239.6
8	1.96	239	2.09	239.2
9	1.02	241	1.12	240.7
10	1.01	240	1.12	240.3

Using t-test to compare the measurements, it is found that the p-values for current and voltage measurements are 0.4755 and 0.8526, respectively. Both values are greater than the significance level of 0.05, indicating that the difference between the manual and prototype measurements is not statistically significant. With these results, the prototype’s measurement results are considered accurate.

The load monitoring system was installed on the utility pole for a week, from March 8 to March 15, to monitor the transformer loading conditions. Current and voltage measurements are taken on the secondary side of the distribution transformer, with the following specifications: 13.2/0.24 kV, 10 kVA, 60 Hz. After testing, it was observed that the peak load of the transformer reached 14.96 kVA at 5:41 PM on March 9. This peak load result shows that the transformer is operating at 149.55% of its rated capacity, indicating that the transformer is overloaded.

Table II shows the recorded daily peak loads of the transformer and the equivalent operating conditions during testing.

Table II. Recorded Transformer Loading

Date	Peak load (kVA)	Percent loading	Operating condition
March 8	14.65	146.50%	Overloaded
March 9	14.96	149.55%	Overloaded
March 10	11.94	119.38%	Overloaded
March 11	14.17	141.69%	Overloaded
March 12	11.89	118.88%	Overloaded
March 13	12.87	128.71%	Overloaded
March 14	14.07	140.66%	Overloaded
March 15	13.16	131.55%	Overloaded

Based on the results shown in Table II, it is evident that the transformer is operating in an overloaded condition and should be subject for uprating. One of the advantages of using the load monitoring system is that the daily transformer loading conditions can be observed remotely using data visualization without the need to go to the site for manual data collection. Fig. 12 shows the data visualization of the recorded data using Google Looker Studio.

Fig. 12. Data visualization using google looker studio: (a) Current measurements during testing, (b) Voltage measurements during testing, (c) Apparent power measurements during testing, and (d) Percent loading of the distribution transformer during testing.

Conclusion

In conclusion, the load monitoring system has been designed to determine the loading conditions of a distribution transformer in real-time and remotely. Since a split core current sensor was used, the load monitoring system can be implemented without the need to de-energize or remove the distribution transformer from operation. Using the system, it was able to determine the loading conditions of the distribution transformer and showed that it is operating in an overloaded condition. As the data is recorded continuously and remotely, manpower is not required to collect information on-site, thus making data collection more efficient.