To minimize coking and maximize the conversion rate in a steam cracker, olefin producers need to closely monitor the furnace’s coil outlet temperature (COT). Various types of sensors measure this key parameter. WIKA USA tested the common sensors designs to evaluate their performance and accuracy.
This article is a continuation of “Improving Olefin Production Efficiency with Coil Outlet Temperature (COT) Measurement.”
Designs for COT sensors
The market currently offers three main types of sensors for measuring temperature at the coil outlet of ethylene steam crackers. Each design has its advantages and disadvantages.
(top) drilled thermowell coated with erosion-resistant alloy, (bottom) drilled bar of erosion-resistant alloy welded to thermowell stem
1. Immersed sensor with thermowell
A replaceable thermocouple is installed within a thermowell that’s inserted directly into the process stream. Since the stream in an ethylene cracker is extremely hot as well as abrasive, due to the presence of hard coke particles, manufacturers protect thermowells in two ways: by coating the outside with an erosion-resistant allow, or by fabricating the tube from a solid piece of an erosion-resistant alloy and welding it directly to a standard thermowell stem.
The advantage of using an immersed temperature sensor is that the thermowell is in direct contact with the process. This design is also easy to install if a process entry is available. The disadvantage is that despite the protective erosion-resistant alloy, immersed thermowells are still constantly buffeted by coke particles and will need to be replaced regularly, with the frequency depending on each facility’s operating conditions and the quality of the feedstock. (Some plants turn their thermowells 180° every six months or so to increase the thermowells’ lifespan.)
2. Permanently welded surface sensor
The sensor block of this tubeskin thermocouple is welded to the surface of a coil outlet pipe.
Tubeskin thermocouples are commonly used for these installations. With this style, the thermocouple is welded directly to the outside of the coil outlet tube. Thermal insulating material covers the sensor block to protect the measuring point from the influence of ambient conditions.
The advantage of this design is the guarantee that the measuring point is in direct contact with the tube surface. And since tubeskin thermocouples never come into contact with the process stream, they typically have a longer lifespan. The disadvantage of a permanently welded sensor is that installation calls for a trained technician using welding procedure specifications (WPS) tailored to the metallurgy of the coil; otherwise, sensor performance – particularly accuracy – is compromised. And when a replacement is needed, the sensor is either cut or abandoned, and a new sensor is permanently installed at another target location.
A replaceable thermocouple is inserted into a curved guide tube whose tip has been welded onto the surface of a furnace coil.
3. Replaceable surface sensor
In this design, a curved guide tube is mounted onto the surface of a coil outlet pipe, with the tip of the guide tube fully welded to the coil. (An incomplete weld can create an air gap, which results in heat loss or increases the temperature influences from ambient conditions.) A thermocouple is inserted into the guide tube, making direct contact with the tip of the tube to achieve full conduction.
The advantage of this design is that when a sensor needs to be replaced, operators can simply remove the failed thermocouple and insert a new one; no additional welding is needed. One disadvantage of this design is that when installing the sensor into the curved guide tube, it is difficult to know for certain whether the sensor extends to the end of the tube. If it does not, the air gap will have a negative impact on the sensor’s accuracy. Similarly, accuracy is compromised if the sensor does not make contact with the tube surface.
Process temperature vs. tubeskin temperature
The industry preference for COT sensor styles are diverse, as immersed-sensor systems and surface-sensor systems each have their advantages and disadvantages. Some operators prefer having data on their process temperature, while others prefer knowing the tube metal temperature.
When deciding on a sensor design for a particular application, operators weigh factors such as initial cost, ease of installation and replaceability, and sensor longevity. However, the primary consideration should be accuracy, followed by a fast response time.
So, how do these styles compare with one another when monitoring temperature at the coil outlet? To find the answer, a team at WIKA USA conducted a series of tests on three main designs.
COT Sensor Test Design: Setup
Testing for these sensors was conducted at WIKA’s R&D Center in Pasadena, Texas, which was purpose-built to be able to replicate real-world applications. Within the R&D Center is a process loop containing a 9.6M BTU fired heater and fully functioning reactor. This setup provides the ability to run repeatable, controlled tests on various products in different stages of a live refining process.
Furnace installations
To run tests on the COT measurement styles, the fired heater was used to simulate COT conditions. While WIKA does not have an ethylene cracker to work with, custom fixtures were fitted onto the heater to allow radiant gases to pass through an exhaust tube. Each sensor style was fitted to its own outlet tube, which included built-in reference sensors for comparisons. The entire system was designed to withstand temperatures as high as 900°C (1650°F).
Sensor styles
The tests aimed to compare common sensors that would be installed in the industry. The sensors tested include:
- Direct-immersed thermowell with replaceable thermocouple
- WIKA V-Pad® original design, permanently welded tubeskin thermocouple
- WIKA V-Pad® version 2.0, permanently welded tubeskin thermocouple
- Permanently welded guide tube in standard design, with replaceable thermocouple
- Permanently welded guide tube in WIKA design, with replaceable thermocouple
Each of the three sensor styles was mounted onto its own pipe, which represents a coil outlet tube, and wrapped with ceramic fiber insulation and a metal cladding shield.
Locations of the internal reference thermocouples (IRTC-X) and surface reference thermocouples (SRTC-X)
COT Sensor Test Design: Reference sensors
Of course, it would be impossible to determine sensor accuracy without knowing what the actual temperatures are. In order to perform the comparisons from one style to another, a set of reference sensors were installed onto each of the outlet tubes under test. These reference sensors included both internal reference thermocouples (IRTC) and surface reference thermocouples (SRTC). The IRTCs were inserted into the center of the outlet tubes to provide reference of the internal flue gas temperatures, while the SRTCs were welded directly to the outer tube surface for tube temperature reference.
Preparation for testing
Each set of sensors was tested over two days, with each day comprising two interval temperatures. For the first few days, the system was run to achieve a benchmark at temperatures of approximately 788°C (1450°F) and 882°C (1620°F) as measured by the lower internal reference sensor (IRTC-1).
The internal reference thermocouples maintained a stable temperature between the bottom and top locations with only a small increase in temperature at the top internal reference thermocouple (IRTC-2). The surface reference sensors, however, showed a temperature gradient with temperatures increasing as the outlet tube extends further away from the exit location of the furnace. Readings level off at the 9″ (229mm) mark. For the purpose of our comparisons, each sensor style was evaluated against the lower internal reference sensor (IRTC-1) as well as the surface reference sensors located at 9″ (229mm), 12″ (305mm), and 15″ (381mm): SRTC-3, SRTC-4, and SRTC-5, respectively.
Comparing the Three COT Sensor Designs
When analyzing the data, all the tested sensors provided readings with a large deviation from the internal reference thermocouple. While the thermocouple readings from inside the thermowell were found to be closest to the internal reference sensor, it showed similar deviations to the other tested sensors at the lower test temperature. In contrast, its performance improved with about a 10% lower deviation at the higher test temperature.
While testing showed similar readings between the immersed sensor and the other sensor styles, it is important to note that different configurations of outlet tube and thermowell sizes will play a factor in an immersed sensor’s accuracy. The deviations from actual process temperature to readings from a thermocouple within a thermowell will vary based on several factors, including but not limited to
- Pipe sizes, which allows for different insertion lengths
- Thermowell diameters and materials
- Sensor diameters
COT design accuracy compared to surface reference sensors
When compared to the average of surface sensors SRTC−3, SRTC−4, and SRTC−5, four of the five tested designs showed accuracies within standard limits of error according to ASTM E230 specifications. The two designs that showed the smallest deviations are
- Permanently welded surface sensor V-Pad 2.0
- Replaceable surface sensor with WIKA-designed guide tube
These two designs showed similarities in readings at the higher test temperature, with the V-Pad® 2.0 providing significantly less deviation at the lower test temperature. The worst performance came from replaceable surface sensor with standard guide tube. It is worth noting that the immersed sensor with thermowell performed almost as well as the V-Pad 2.0 and the WIKA-designed guide tube. As mentioned earlier, the design of the thermowell – its material, dimension, immersion length – as well as the dimensions of the process pipe, are all factors in its temperature sensor’s accuracy.
Sensor response times
When recording the temperature readings of the five test sensors, we also tracked their response times. However, since each style of sensor was installed onto different outlet tubes and tested on different days, slight variations in ramp-up times and temperatures are present in the response time data. As a result, the response times of each design cannot be compared directly against one another, but instead are overlayed to give a relative comparison for analysis.
Data show that the immersed thermowell provided the quickest response time, with the two V-Pads being similar to one another and next fastest. When looking at the guide tube designs, there is a considerable difference between the standard design and WIKA design, as the WIKA design has similar performance to the V-Pad designs.
Findings, Next Steps, and Recommendations
After detailed analyses of the test results, we came up with three main findings.
1. Temperature sensing with thermowell showed greater-than-expected deviation from reference sensors. One of the surprising outcomes of the tests was the high amount of temperature deviation when using a thermowell. We had anticipated that the immersed test sensor would deviate the most from the surface reference sensors, but that it would be much closer to the internal reference sensor than what the results showed. One thought is that the amount of thermal mass on the outside of the process was larger than on the inside, which affected the overall temperature readings of the immersed test sensor. Further testing or computational fluid dynamics (CFD) analyses would be needed to prove this hypothesis.
2. Insulation plays a critical role in sensor readings.
Not surprisingly, ambient conditions had a large impact on the surface temperature when comparing readings of reference sensors with and without insulation. What might be slightly more surprising is that uninsulated tubes do not have a consistent temperature over its entire length due to the effects of ambient temperatures, the furnace roof, or both. Determining the optimal type and amount of insulation, which would require additional testing, might be interesting for some ethylene producers.
3. Uninsulated fixtures have a negative effect on sensor readings.
Uninsulated parts of the furnace, such as exposed fixtures, had an effect on tube temperatures. From reviewing the data from all surface reference sensors, we see that temperatures started to level off at 9″ from the furnace roof and remained consistent up to 18″ away – the location of the last surface reference sensor in our tests. When determining the installation location of any style of sensors, WIKA recommends keeping a minimum distance of 9″ away from any uninsulated fixture that is connected to the tubes being measured.
Conclusions and Next Steps
The petrochemical industry realizes the importance of monitoring coil outlet temperature for olefin production efficiency, and there are several options available in the market for measuring this critical parameter. Purchase price, of course, is a consideration, and that’s easy to see. What’s less obvious is which of the options offers the greatest accuracy and repeatability. A review of the available literature did not reveal much regarding which sensor design was optimal, which is one reason we were eager to carry out these tests.
With the testing performed, it is clear that some sensor designs have smaller deviations than others at certain operating conditions. While there is no-one-size-fits-all sensor design, considerations can be made to better understand which sensor type is ideal for each installation. Take, for example, the measurements of surface temperature. During the tests, the V-Pad® version 2.0 provided unmatched accuracy for surface measurements for outlet tubes operating at the lower temperatures. On the other hand, deviations of the WIKA guide-tube design and V-Pad® version 2.0 were similar at the higher test temperature. Having the ability to replace the thermocouple in the guide tube design could be an added benefit that some operators prefer without having to trade off in their accuracies.
When looking at the immersed sensor designs, depending on the installation configuration and size of the outlet tubes, there may not be much of an accuracy improvement compared to one of the surface measurement options. In this case, an evaluation should be made to determine if the added maintenance of an immersed sensor is worthwhile or if a surface measurement option is more desirable.
This initial series of tests covered only a small sample of sensors and insulation configurations available. Future tests, by WIKA USA and others, would build upon these results and contribute to the body of knowledge on which of the various sensor designs offer the highest accuracy, best repeatability, and fastest response times (based on specific application and installation details) – all factors in improving process performance.
Contact Alina Green for more information about how WIKA USA can help increase run times, boost conversion rates, and improve fuel efficiency in your olefin production facility.