Gas turbine hot sections condition monitoring based on operational thermal dataset

Abstract

This paper aimed to monitor the exhaust gas temperature (EGT) from the end of the low-pressure turbine (LPT) of a gas turbine for period of 6 months. To achieve this, 16 thermocouples were strategically placed to gather data at different points in the exhaust system. This comprehensive approach allowed for a detailed analysis of the exhaust gas temperature, which is a critical factor in the health of hot section of gas turbines. The results of this study provide valuable insights that can be used to optimize the periodic inspections of gas turbines and improve their decisions.

The investigation of thermal fluctuations that can cause damage to hot components has been carried out using two statistical methods - Standard deviation and Skewness. By analyzing the standard deviation, the degree to which the temperature values vary from the mean and relative normal condition of each unit can be determined. Skewness helps to identify whether the temperature data is skewed towards the high or low values, indicating the presence of potential anomalies. The application of these statistical methods is aimed at understanding the impact of temperature fluctuations on hot components and developing maintenance strategies to mitigate their effects.

In order to verify the accuracy of the statistical results, a thorough borescope inspection of the gas turbine is carried out in accordance with the maintenance manual. These inspections were conducted at three distinct intervals to ensure a comprehensive evaluation of the gas turbine condition performance and condition. The results of this inspection serve as a critical component in determining the optimal maintenance and repair plan for the gas turbine.

Specifications Table

Subject	Mechanical Engineering
Specific subject area	Data Mining and Statistical Analysis.
Type of data	Raw Analyzed Filtered
Data collection	The exhaust gas temperature measurements were obtained using 16 thermocouples on LPT exhaust of the gas turbine. The average values of these thermocouples have been recorded in a comma separated values file format (CSV) and converted to .xlsx format. The data is recorded automatically at regular intervals of 500 milliseconds for each thermocouple. This data collection process was carried out daily (24 h) for 6 months. As the data analysis is aimed at predicting damages in hot sections (Liners) of the gas turbine during operation mode, calculations have only been made using data above 700 °C. However, as the size of the files began to increase, it was decided to change the interval for data recording from 500 milliseconds to one minute. This change helped to reduce the size of the files and made it easier to manage and analyze the data effectively without reducing the accuracy of the data.
Data source location	The data was collected at the industrial plants monitoring department at TUGA company, MAPNA group, and Iran.
Data accessibility	Repository name: LPT EGT Trend Data identification number (doi): 10.17632/5mjg4yvt45.1 Direct URL to data: https://data.mendeley.com/datasets/5mjg4yvt45/1

1. Value of the Data

• •For a complicated system like gas turbines, Exhaust Gas Temperature (EGT) plays a big role in machine's operating condition [1], it indicates how the system is operating and where the performance is heading.
• •These data can be used by others in the industry to specify a criteria for the gas turbine Health monitoring.
• •A model can be created from the exhaust temperatures obtained from the thermocouple placed on the gas turbine exhaust section.
• •By processing this type of data, more information can now be extracted from available sources and use it to determine machine health and likely failures.
• •Maintenance schedules can be optimized and intervals based on the model's feedback to reducing the effect of failures. This significantly affect to reduce costs and outages period time.

2. Background

The EGT is one of the most important factors in a gas turbine due to its direct impact on the performance of this system. Some information can be extracted from this dataset by statistical means which can be used in favor of fault diagnosis. In this study two statistical indexes, i.e. standard deviation and skewness [2] have been used to process the data and the output have been monitored for the period of 6 months.

3. Data Description

The data article presents the datasets on the gas turbine studied to analyze LPT EGT with the aim of predicting the damage in hot components of gas turbine according to the operation of the machine. The file “LPT EGT 15apr to 22sep.xlsx" contains the average value of 16 thermocouples of EGT from 15 April 2022 to 22 September 2022 for the period of 6 months and the analysis will be carried on this dataset. Part of the data in the mentioned Excel file is shown in Table 1.

Table 1.

A part of dataset.

Row	Date	Date+Time	Time	Mean value of LPT EGT (°C) ValueY
1	15-Apr-22	15-Apr-22	10:49	27.09314816
2	15-Apr-22	15-Apr-22	10:50	27.19012876
3	15-Apr-22	15-Apr-22	10:51	27.25958274
4	15-Apr-22	15-Apr-22	10:52	30.36729175
5	15-Apr-22	15-Apr-22	10:53	247.8865097
…	…	…	…	…
230,322	22-Sep-22	22-Sep-22	23:59	729.8986909

4. Experimental Design, Materials and Methods

The exhaust system of a gas turbine is monitored using 16 thermocouples placed at various points. The thermocouple specifications have been described in Table 2.

Table 2.

The thermocouple specifications.

Type	Connection Type	Material	Well Dimension (mm)	Temperature Range (℃)
K	Internal Thread	Platinum	80	0–1300

These thermocouples measure the data which is then sent to the Junction Box (JB). From there, the data is transmitted via cables and terminals to a Human-Machine Interface (HMI) located in the control room. The HMI displays the data collected every second, allowing for easy monitoring of the gas turbine's performance. Fig. 1 provides a visual representation of this process.

Fig. 1.

Data extraction process from thermocouples installed on gas turbine.

To assess the health of the hot sections of a gas turbine, including the liners, and stator blades, we conducted an analysis over six months using 16 thermocouples, as depicted in Fig. 2. This analysis was aimed at evaluating the trend of the mean Low-Pressure Turbine Exhaust Gas Temperature (LPT EGT). Based on the information presented in Fig. 3, it is evident that during normal operation, temperature fluctuations of around 30 °C were observed.

Fig. 2.

Mean LPT exhaust gas temperature trend for 6 months.

Fig. 3.

Mean LPT EGT trend during the operation of the turbine during six months.

The index information for each month is listed separately in Table 3.

Table 3.

Gas turbine temperature indexes in 6 different months.

Skewness	Standard Deviation	Maximum	Mean	Minimum	Skewness
April	700.85	729.89	733.33	0.79	−9.58
May	700.24	729.13	736.14	3.47	−4.85
June	703.10	723.45	731.37	4.76	0.63
July	700.21	719.69	722.98	1.69	−6.48
August	706.61	728.24	732.83	3.81	−1.71
September	702.52	729.89	752.75	3.29	−5.21

4.1. Standard deviation

Based on the data presented in Table 1, it was observed that temperature variations and shifts in temperature ranges varied across different months. During the six-month evaluation period, June, August, and May exhibited higher levels of temperature fluctuations compared to other months, with standard deviations of 4.76, 3.81, and 3.47 respectively. On the other hand, April and July showed the least amount of temperature variation, with standard deviations of 0.79 and 1.69.

4.2. Skewness

Based on the findings, skewness was predominantly negative throughout most of the period, indicating a concentration of temperature data. Specifically, skewness values of −9.58, −6.48, and −5.21 were observed in April, July, and September, respectively. Calculations reveal that the only positive skewness of 0.63 was recorded in July. This suggests that in July, unlike the other months, the concentration of data was observed between 728 and 735 °C, not centering around 730 °C.

Low-temperature fluctuations are beneficial in the hot sections of a gas turbine. The findings indicate that when the skewness value is either positive or negative (significantly different from zero) and the standard deviation is near zero, the temperature fluctuations are minimal. As indicated in Fig. 4, May had a standard deviation of 3.47 and a skewness of −4.85, showing no data concentration and a high dispersion. Conversely, April had a standard deviation of 0.69 and a negative skewness of −9.58, leading to low fluctuations compared to other months, as shown in Fig. 5. However, in the subsequent two months, these temperature fluctuations increased. This trend contradicts expectations, with higher temperature fluctuations observed in various months over the 6-month period. These fluctuations exhibited a periodic pattern throughout the 6-month period. If this pattern persists, due to the increased thermal stresses on the components of the gas turbine's hot sections, potential damage can be anticipated. To assess the health of these areas, it is essential to plan a visual inspection, including borescope checks, at specific intervals during the operation phase.

Fig. 4.

Gas turbine mean LPT exhaust gas temperature trend in May.

Fig. 5.

Gas turbine mean LPT exhaust gas temperature trend in April.

4.3. Experimental analysis

Given the critical importance of maintaining the health of the hot sections of a gas turbine during operation, itʼs essential to closely monitor key areas. The areas where the borescope inspection is performed are indicated in Fig. 6, including liners and stator blades of the high-pressure turbine. As detailed in the previous section, we observed significant temperature fluctuations in the low-pressure and high-pressure turbine (LPT and HPT) blades and their coatings, as well as the liners, over a 6-month period using 16 thermocouples. These observations utilized standard deviation and skewness indicators to gauge the extent of temperature variations. Given these fluctuations, it's anticipated that if the current trend continues, the blades, their coatings, and the turbine liners could suffer damage. To corroborate the findings from the previous section, we planned to conduct borescope inspections at three intervals: at 2000, 4000, and 6000 h of gas turbine operation according to Fig. 7.

Fig. 6.

Borescope inspected areas (liners and stator blades of HPT).

Fig. 7.

Borescope inspection of the gas turbine liners and stator blades of HPT.

The video borescope used in this specific application is equipped with a small yet powerful video camera attached to a flexible tube. The tube is inserted into the liner area and stator blades of the HPT after disassembling burners, as shown in Fig. 7. This enables the inspector to perform a thorough inspection and look for any signs of damage or wear and tear. To aid in the inspection process, the entrance tube's end is fitted with a bright light that illuminates the area, enabling the operator to capture clear, high-quality images or videos of hard-to-see areas. Fig. 8, Fig. 9, Fig. 10 were mentioned in this case to examine the condition of these areas for any visual damage that may have occurred.

Fig. 8.

A 2000-h borescope inspection.

Fig. 9.

A 4000-h borescope inspection.

Fig. 10.

A 6000-h borescope inspection.

Fig. 8 illustrates a segment of the HPT stator blade during the 2000-h borescope inspection.

As it is clear, a small part of the HPT stator blades coating has been damaged. Similarly, according to Fig. 9, in the 4000-h

According to Fig. 10, in the 6000-h borescope inspection, as expected, wider areas of liners and stator vanes have been damaged.

Based on the images obtained from the borescope inspections carried out at three different intervals, it has been determined that the damage on the hot components of gas turbine is related only to the damage on the coating of stator blades of HPT, caused by temperature fluctuations of the gases exiting the turbine. During these inspections, no other types of failures were observed.

As depicted in Fig. 11, the standard deviation has increased over the 6-month evaluation period, while skewness has transitioned from a positive to a negative value, indicating a shift in the distribution of temperature data. This shift suggests an increase in temperature data dispersion and a corresponding rise in the operating temperature of the turbine from 700 °C to 730 °C. Consequently, the concentration of temperature data has significantly diminished.

Fig. 11.

Investigating the number of failures of hot turbine sections according to standard deviation and skewness indices.

According to Fig. 11, three cases of failures have been observed in the 2000-h borescope inspection. The number of defects increased with the increase in temperature fluctuations and reached five items within 4000 h of the inspection. In the end, with the same conditions remaining, seven cases of failure occurred in the 6000-h inspection.

Limitations

Not applicable.

Ethics Statement

No human or animal subjects were involved in this work, and the authors confirm that there are no known conflicts of interest or significant financial support that could have influenced its outcome. The authors have taken precautions to protect any associated intellectual property.

CRediT Author Statement

Masoud Khoshghiafehgan: Data curation, investigation, methodology, supervision, validation, writing-original draft; Amir Akhlaghi: Data curation, Data analysis, investigation, methodology, validation, writing-original draft, writing-review and editing; Abbas Ranji: writing-original draft, writing-review and editing; Majid Yeganeh Mazhar: writing-original draft, writing-review and editing.

Data Availability

• LPT EGT Trend (Original data).