Thermal Behavior and Power Efficiency Comparison of AC vs. DC Electrical Heating in a Distillation Column Using Infrared Thermography Analysis

Heat exchange systems are fundamental devices in chemical engineering, used to facilitate the transfer of thermal energy between two process streams, even without direct physical contact. Their incorporation into equipment such as distillation columns allows for optimizing energy consumption, improving component separation efficiency, and maintaining stable operating conditions through precise control of temperature profiles¹.

A power actuator in a distillation column is considered a heat exchange system, composed of an electrical resistance that, in both direct current (DC) and alternating current (AC), obeys Ohm's law, establishing a direct relationship between voltage, current, and resistance. However, the behavior of the resistor when using AC is more complex due to the influence of frequency, impedance, and the inductive and capacitive effects of the system. However, it offers certain advantages in specific industrial applications^2,3.

The behavior of a real resistor differs from its ideal because physical materials can induce phenomena such as parasitic inductance and, in certain cases, parasitic capacitance⁴. These effects depend on the AC signal frequency, being distinguished in terms of frequency values. In one case, the reactive effects (inductive or capacitive) are practically negligible; therefore, the resistor behaves linearly and obeys Ohm's law almost perfectly. Thus, its response is comparable to that of an ideal resistor used in DC, when considering low frequencies.

For high frequencies, the actual resistance begins to exhibit an inductive component, especially if its construction includes elements such as windings or long connections⁵. The improper use of electrical resistors in distillation processes can cause thermal shocks in the column boiler, especially when the equilibrium of the thermosiphon effect is altered, a phenomenon that occurs due to the interaction between heat transfer and the natural flow of the liquid induced by density differences⁶.

The heat transfer rate in the boiler is determined by the effective length of the heating and evaporation zones, as well as by the geometric design of the exchanger⁷. Furthermore, this rate depends significantly on the operating pressure and vapor content, since the heat transfer coefficient in the evaporation zone is considerably higher than in the preheating zone⁸.

This work proposes a comparative study between AC and DC power supplies to feed a heating resistor inside a distillation column, evaluating their impact on the thermal efficiency of the process, temperature profile in the boiler, energy consumption, and stability during the distillation cycle. The main characteristics of the plant are: a 2 L boiler plate, a 300 W heating resistor, bottom product extraction, a double spiral condenser, a L reservoir tank for distilled products, and on-off reflux valve.

1. Configuration and procedures for mixture preparation, process execution from the local interface, and power supply startup

1. Preparation of the mixture to be used in the process inside the boiler (Figure 1).
   1. Place 1 L of ethanol and 1 L of distilled water in separate flasks.
   2. Place the contents of each flask inside the plate boiler plate.
   3. Ensure the instrumentation is correctly positioned.
      1. Check the hose connections on the condenser from the cooling water pump.
      2. Check the placement of the sensors on the plates of the distillation column body.
      3. Check the conditions of the cooling water pump actuators, heating resistor, and reflux actuator.
2. Execute the local interface on the PC and data acquisition card (Figure 2).
   1. On the local PC, open the "distillation column" software.
   2. Connect the data acquisition card to the local PC.
   3. Verify communication between the data acquisition card and the local PC.
      1. Validate the items in the pop-up window.
   4. Select the mixture to be used in the local interface.
   5. Start the process from the local interface by clicking the Start button.
      1. Verify receipt of information from local sensors in each of the interface windows on the local PC.
         NOTE: Although the experiment is limited to a binary mixture of ethanol and water, the interface allows for the observation of different mixtures. Since heat transfer between the RTD and the liquid depends on general thermal properties such as heat capacity and conductivity, it can be assumed that the methodology employed, which relies on experimental data to define thermal behavior, is valid for other mixtures under controlled conditions. This approach is supported by studies on distillation columns that dynamically model ideal binary mixtures^9,10,11.
3. Preparing for AC or DC power supplies to power the heating resistor.
   1. Using a multimeter as a voltmeter, set the output voltage value from the front panel of the power supply, as appropriate (AC or DC). Once configured, turn off the power supply.
   2. Connect the voltage supply to the resistance thermometer installed in the boiler plate.
   3. Energize the heating resistor by turning on the power supply.
   4. Verify current and voltage monitoring from the local interface.
      NOTE: Table 1 indicates the voltage and current values used for the experiment for both AC and DC.

2. Setup, acquisition, and analysis of thermograms and temperature data on the in-process boiler plate

1. Setup and acquisition of thermograms using the in-process boiler plate thermal camera.
   NOTE: For the experiment, operating hours and measurements were performed from 1 to 3 PM, when the ambient temperature inside the laboratory was 22° to 24.5 °C.
   1. Select the Iron color palette from the thermal camera menu.
   2. Set the emissivity correction value to 0.95 from the thermal camera menu.
      NOTE: The camera can be self-calibrated and is periodically calibrated by trained personnel. The emissivity value is selected based on the roughness, thickness, and transparency characteristics of the glass that makes up the boiler tank. Therefore, an emissivity value of 0.95 was considered for the thermal measurement¹³. Since the difference is minimal, the impact on the temperature estimate is negligible within the experiment's operating range.
   3. Position the thermal camera approximately 80 cm away from the boiler plate.
      NOTE: The measurement in 2.1.3 was considered based on the camera's minimum operating distance (0.6 m) due to its resolution (140 x 140 pixels) and field of view (29° x 29°)¹², therefore, the distance used allows capturing relevant thermal details without compromising the pressure measurement on the boiler plate.
   4. Holding the thermal camera approximately 60 cm above the floor, locate the camera pointing toward the boiler plate in the distillation column.
      NOTE: The measurement in 2.1.4 was selected based on the height of the plate boiler.
   5. Press the save button to record the thermogram to the thermal camera's memory.
      NOTE: Because the transient state is slower depending on the decrease in the supply voltage applied to the heating resistor, the action in step 2.1.5. can be performed at 20 to 30, 50, or 60 s for 20 V (AC and DC).
      1. Capture every 5 s during samples 1 to 1500 indicated on the local interface, repeat for 20, 60, and 100 V for both AC and DC.
      2. For 100 V (AC and DC), capture every 10 s during samples 1500 to 2800 indicated on the local interface, for 20 V (AC and DC), capture during the sample interval 1500 to 5500, and for 60 V (AC and DC), capture every 15 s during samples 1500 to 4000.
      3. For 100 V (AC and DC), capture every 20 s from sample 2800 until steady state is reached, for 60 V (AC and DC), capture every 30 s from sample 4000 until steady state is reached, and for 20 V (AC and DC), capture every 20 s from sample 5500 until steady state is reached.
   6. Record the name of the captured thermogram and the value of the boiler plate temperature sample displayed in the local interface.
2. Acquisition of boiler plate temperature data through the local process interface.
   NOTE: Figure 3D,E show the process mentioned in 2.2 in detail.
   1. Store the temporary files generated by the interface.
   2. Rename and save the files in the current process folder.
   3. Collect the .csv files generated during the test and generate a single file with all the data obtained from the process.
      NOTE: The interface stores process data every 1020 samples, which is equivalent to one section of the process. These are temporary files that must be stored for later processing. 0.5 s is the sampling time for each sample. The saved numerical data corresponds to the dynamic behavior of the process generated by a temperature sensor measuring the vapor inside the boiler.
3. Preparation of boiler plate temperature data and captured process thermograms.
   NOTE: Figures 3A,B,C,F,G show the process mentioned in 2.3.
   1. Extract thermograms from the thermal camera.
      NOTE: The thermograms show the temperature distribution resulting from the RTD heating according to the voltage used.
   2. Review the thermograms and discard any unwanted ones.
   3. Manually mark a reference point on the thermograms.
   4. Create the process temperature overview graph using the process data file generated in 2.2.3.
   5. Relate the boiler temperature data from the local interface to the captured process thermograms.

3. Analysis and processing of thermograms for the boiler plate process

1. Development of an algorithm for thermogram processing.
   1. Preparation of graphs for visualizing average trends.
      1. Load the image into a variable.
      2. Define the size of the vectors for the graph axes, using the image variable.
      3. Define the area for detecting minimum and maximum temperature ranges based on the analysis image.
      4. Prepare the figure for graphing.
2. Processing thermograms (Figure 4).
   1. Load the image into a variable.
   2. Define a search area for the reference point marked in 2.3.3.
   3. Isolate the reference point and obtain its coordinate.
   4. Indicate the analysis area using the coordinates in 3.2.3.
   5. Convert the analysis area to grayscale.
   6. Calculate the mean of the area defined by 3.2.5.
   7. Isolate the maximum and minimum temperature areas in the image.
   8. Convert the graphical numbers to numeric text.
   9. Convert the mean values to scaled maximum and minimum temperatures according to the image ranges.
   10. Store the total mean values and labels.
   11. Plot the total mean results.

The results are from six tests developed with the values mentioned in Table 1. Thermal information was considered for the analysis, as well as data obtained through the local interface.

Comparative analysis for a 100 V AC-DC supply voltage
The data obtained from the process using the local interface indicated that for a DC power supply, the temperature measurement varies throughout the process, while using AC (Figure 5A), the measurement is more stable.

Notably, the time to reach the transient state in the process is the same for both cases (approximately 6000 samples). However, there is a variation in the thermal evolution within the boiler plate, which is observable in Figure 6A for VDC and Figure 6C for VAC.

The minimum and maximum temperature values for each thermogram were obtained to observe the points with the highest and lowest temperatures. Additionally, the mean of the thermal data was obtained to observe the average temperature in the region of interest. Finally, the coefficient of variation was obtained to compare the dispersion between different areas. The statistical results of the isolated section for each thermogram (Figure 4E) are presented in Table 2 for 100 VDC and in Table 3 for 100 VAC.

From the coefficient of variation, it can be concluded that using VDC presented more controlled areas due to a smoother progression and increasing but not abrupt dispersion (Figure 6B). On the other hand, using VAC, more jumps were observed (Figure 6D), which may indicate greater exposure to thermal gradients. Figure 6E presents a graphical comparison of the results.

Some of the thermograms used to plot the average intensity per column in the analysis region are presented in Figure 7B and Figure 8B in the same order.

A total of 105 and 110 process captures were considered during the test. An example of sample capture distribution is shown in Figure 7A and Figure 8A, following the same order.

Comparative analysis for a 60 V AC-DC power supply
When comparing the temperature response values given by the sensor installed in the boiler plate for VAC and VDC values, the temperature measurement varies between 0.5 °C and 1 °C, compared to the use of a VAC power supply (Figure 5B), which has a more stable measurement. When observing the thermal evolution given by both power supplies, there is a clear upward trend in temperature in the plate boiler when powered by VAC, showing a gradual increase during the start-up phase of the process (Figure 9B).

In contrast, if DC is used, the thermal evolution is much more pronounced, as evidenced by the faster heating of the mixture (Figure 10B), compared to that observed with VAC.

As part of the previous analysis, the average of the columns was calculated for the area under observation, obtaining the average intensity value, which was normalized within the maximum and minimum ranges of the process during the test. The coefficient of variation was similar to that observed for 100 V AC and DC. The statistical results are presented in Table 4 for 60 VDC and Table 5 for 60 VAC.

Figure 9A and Figure 10A show, from the local interface data, the time of thermogram capture during the test for AC and VDC.

Only a small number of samples were used for this test due to the prior knowledge that the low-power process tends to take longer to reach steady state. The data obtained showed the expected behavior based on the information collected.

Comparative analysis for a 20 V AC-DC supply voltage
The graphical temperature trend behavior at the local interface is similar to that observed in the previous tests (Figure 5C). The presence of variation is noticeable for the VDC test.

If the thermal evolution is observed using VAC, it remains very gradual and stable (Figure 11A). The thermograms of the graphical trend of the local interface product were considered as shown in Figure 11B.

Figure 5D shows a comparison of the temperature evolution under different AC and DC power supply conditions during the test.

In conclusion, the coefficient of variation can be used as a statistical indicator of how thermal energy is distributed in a system. When a high coefficient of variation is present, it may be the result of large thermal gradients, which implies heat flow (Joule transfer between zones). On the other hand, in processes, a low variation coefficient indicates that thermal energy has been distributed efficiently.

Figure 1: Process preparation methodology. (A) Prepare 1 L of ethanol and 1 L of distilled water. (B) Place it inside the boiler plate. (C) Check hose connections (blue), temperature sensors (green), connectors (yellow), and the installed heating resistor (red). Please click here to view a larger version of this figure.

Figure 2: Local interface execution methodology. Indicates the process execution methodology from the local interface. (A) Program icon. (B) Connection of the data acquisition card to the PC. (C) Serial monitor items indicate a successful connection. (D) Mixture menu, for selecting the mixture to be used. (E) Mixture selection. (F) Execution icon. (G) Successful process execution. Please click here to view a larger version of this figure.

Figure 3: Data and thermogram preparation. (A) Extraction of thermograms. (B) Selection of thermograms. (C) Setting a reference point for processing. (D) Data generation from a local PC. (E) Data collection into a single file. (F) Process temperature graph using CSV files. (G) Relationship of thermograms with the boiler plate temperature line. Please click here to view a larger version of this figure.

Figure 4: Selection process for the analysis area. (A) Thermogram under study. (B) Search area. (C) Search area isolation, (D) Reference point isolation and coordinate extraction, and (E) Area definition. (F) Range search area. (G) Isolation of temperature ranges for conversion to numeric text. (H) Normalization of the intensity vector with maximum and minimum temperature values. (I) Column average graph of the area under analysis with a temperature range applied to each analyzed image. Please click here to view a larger version of this figure.

Figure 5: Process temperature comparison in the boiler plate. (A) Graphical comparison for 100 V AC and DC. (B) Graphical comparison for 60 V AC and DC. (C) Graphical comparison for 20 V AC and DC. (D) Comparison of temperature evolution under AC and DC power supply conditions. Please click here to view a larger version of this figure.

Figure 6: Statistical graphs. (A) 100 V DC graph of area column averages under analysis with an applied temperature range. (B) Coefficient of variation graph for 100 V DC. (C) 100 V AC graph of area column averages under analysis with an applied temperature range. (D) Coefficient of variation graph for 100 V AC. (E) Comparison of AC and DC coefficient of variation graphs. Please click here to view a larger version of this figure.

Figure 7: Trend relationship with thermogram capture for 100 V DC. (A) Thermogram label associated with the trend. (B) Some thermograms captured during the process. Please click here to view a larger version of this figure.

Figure 8: Trend relationship with thermogram capture for 100 V AC. (A) Thermogram label associated with the trend. (B) Some thermograms captured during the process. Please click here to view a larger version of this figure.

Figure 9: Trend relationship with thermogram capture for 60 V AC. (A) Thermogram label associated with the trend. (B) Some thermograms captured during the process. Please click here to view a larger version of this figure.

Figure 10: Trend relationship with thermogram capture for 60 V DC. (A) Thermogram label associated with the trend. (B) Some thermograms captured during the process. Please click here to view a larger version of this figure.

Figure 11: Trend relationship with thermogram capture for 20 V DC. (A) Thermogram label associated with the trend. (B) Some thermograms captured during the process. Please click here to view a larger version of this figure.

Figure 12: Effect of course on mean data within the analysis. (A) Thermogram under analysis. (B) Isolated section for analysis. (C) Average mean chart with red-marked area indicating the effect of the cursor on the trend. Please click here to view a larger version of this figure.

Table 1: AC and DC voltage values used for the test. Please click here to download this Table.

Table 2: Minimum and maximum temperature, mean, and coefficient of variation statistical results for thermograms obtained from the 100 V DC test. Please click here to download this Table.

Table 3: Minimum and maximum temperature, mean, and coefficient of variation statistical results for thermograms obtained from the 100 V AC test. Please click here to download this Table.

Table 4: Minimum and maximum statistical results for temperature, mean, and coefficient of variation for thermograms obtained from the 60 V AC test. Please click here to download this Table.

Table 5: Minimum and maximum statistical results for temperature, mean, and coefficient of variation for thermograms obtained from the 60 V AC test. Please click here to download this Table.

The procedure presented provides a general idea of the importance of correctly selecting the feed for a chemical process such as a distillation column; however, the process has some critical points that must be addressed with great care to avoid unfavorable results.

When capturing thermograms, it is important to consider using a tripod to locate the camera. This prevents accidental movement while capturing thermograms.

Files generated by the local interface must be stored or copied to the general file before being deleted, as this could result in data loss.

It is important to consider the temperature range presented in each thermogram, as this could result in the loss of the trend of increasing heating in the boiler plate.

For the detection of temperature ranges in each image (as mentioned in 3.2.7), it is extremely important to define the numerical detection zone fully visible. During the design, it was observed that if this zone is not considered, it can produce values different from those presented in the thermograms.

Within the analysis of the thermograms presented, it is possible to observe a cursor, which, in the analysis of averages, this cursor is represented in the graphic response as a change in temperature, as shown in Figure 12. The authors are aware of this change; therefore, this was taken into account during the test.

To avoid losing the information generated from the local interface, an additional program that detects file generation, counts the data content, and exports the data to an Excel document to store the information as the process is active can be developed.

It would be advisable to establish a connection between the thermal camera and the local PC to monitor the process, sharing not only images but also real-time video thermograms of the process. This would allow for a more direct relationship between the data from the process sensors and the thermograms captured by the thermal camera video.

The aim is to use this information to perform a fatigue test on the thermal actuator in order to evaluate its continuous wear and how this affects the transient state of the process. In addition, a thermogram estimation model will be developed to predict future behavior according to given conditions, and a fault detection system will be implemented using the thermograms.

The limitations of this experiment are mentioned below. The laboratory does not have ambient temperature control, so a scheduled operation was considered. However, ambient temperature was a variable factor during the tests. For the 100 V AC and 100 V DC tests, due to the rapid heating rate, some thermograms were not captured correctly, even with care. Finally, taking the thermogram as a time series provides a discretized (non-continuous) view of the temperature change on the boiler plate.