SIRG/Collaborative Research: Distributed Subwavelength Photonic Sensors for In-situ High Spatial and Temporal Resolution Monitoring in Manufacturing Environment

B.G. Thomas, Michael Okelman, Joydeep Sengupta, Claudio Ojeda, Gogi Lee

National Science Foundation SIRG DMI # 05-28668 and Continuous Casting Consortium

Quick Links:

Project Overview
National Science Foundation Support
Industry Support
Summary of Activities (Research & Educational Activities, Experimental Studies, Modeling Studies)
Overview of Findings
Training and Development
Outreach Activities
NSF Publications Describing This Work
Journal Publications
Book(s) or other one-time publication(s)
Contributions

Project Overview

Monitoring of mold level and meniscus behavior is important for controlling quality during the continuous casting process.  This project aims to develop new sensors to measure temperature in the mold very near to the meniscus, initially to use as a new research tool to investigate meniscus behavior to better understand defect formation.  The ultimate goal is to revolutionize online thermal monitoring of industrial continuous casting molds.  A process will be developed to insert sensors manufactured at UW Madison into the mold coating layer.  Tests of sensor integrity will be conducted, data collected, and the signals analyzed using computational models.  The meniscus region will further be modeled computationally to predict events during an oscillation cycle – including modeling of the sensor itself. This will determine the relationship between the sensor signal and the actual meniscus events.  Insights gained will enable optimization of the size and location of the new sensors and interpretation of their signals to gain maximum benefit from their installation into operating molds.

Return to Top

National Science Foundation Support:

Collaborators:
Xiaochun Li, ME, University of Wisconsin - Madison
Chee Wei Wong, ME, Columbia University
David A. Dornfeld, ME, University of California - Berkeley
Hongrui Jiang, ECE, University of Wisconsin - Madison

October 1 , 2005 to September 30, 2008
NSF Division: 05-526, "Sensors and Sensor Networks (Sensors)"
NSF Program Director: Abhijit V. Deshmukh, (703) 292-8330
NSF Grants Official: Maria Valerio, (703) 292-8212

Return to Top

Industry Support

Sumitec Seimens (VAI Services): Facilities
Sumitec (VAI Services) of Benton Harbor, Michigan runs a refurbishing and electroplating facility for continuous casting molds. Mike Powers of Sumitec has been helping us to test our designs using their real-world facility. This will avoid problems with scale-up or other differences between lab-scale experiments and full commercial implementation of the new sensor technology we are developing.

We have reached agreement with Nucor Steel, Decatur, Alabama, to test our new sensor in service once the design is proven in trials at Sumitec. Ron O'Malley is the Nucor contact heading this planned effort.

Return to Top

Summary of Activities

Research and Education Activities:
Work at the University of Illinois has focussed on implementing sensors into the continuous casting mold used at the steel plant. This has involved experimental investigation of 1) the attachment of glass fiber and nickel-based sensors onto the copper mold surface, and 2) ability to embed the sensors into the mold coating during the eletro-plating process, and computational modeling investigations of 1) heat-transfer consequences of imperfect plating to determine standards for success, and 2) mechanisms of initial solidification including heat extraction during oscillation mark formation, in order to extract useful results from the finished sensors in the future.

Many of the problems in continuous casting arise during the initial solidification at the meniscus in the mold, including defects in the final cast product as well as cracks in the mold surface due to thermal stress. These problems dictate many aspects of steel quality and productivity. Although the
mold hot face is an ideal location to monitor, it presents a very hostile environment, with copper mold surface temperature ranges from 200 to 400oC, with instantaneous spikes that might reach 800oC during a mold level fluctuation. In current industrial practice, a large number of thermocouples are used in one continuous casting mold, destructively inserted into mold through channels drilled in copper mold, far from the hot face due to safety considerations. This limits their usefulness, as their response time is too slow to capture the rapid events that occur at the meniscus due to the dampening of the temperature signal caused by the thick copper mold between the solidifying steel and the thermocouple. Sticker breakout detection systems, installed in almost every caster worldwide, use thermocouple signals interpreted by control systems to take corrective action, but are limited by inadequate sensor technology and insufficient understanding of how to interpret the signals.

The proposed sensors have a temperature accuracy of +/- 2oC and a 0.01s response time, capable of capturing temperature fluctuations of greater than 100oC in 0.1s, much more sensitive than conventional thermocouples, and are designed to measure heat flux and temperature within 1 mm of the meniscus to more accurately predict level fluctuations and quality problems, as well as providing additional insights into the casting process. Improved versions of the sensor strip design and manufacturing process are being developed through collaborations between the University of Illinois, where the sensor design, modeling, testing, evaluation, and implementation is being coordinated, the University of Wisconsin-Madison, where the sensor strip will be fabricated, a commercial mold manufacturer, where the strip will be attached to a commercial mold, and a commercial steel company, where the mold and sensor will be tested in service.

In sum, this project aims to develop and validate a new type of in-mold sensor for use in the commercial continuous casting of steel. The aim is 1) to revolutionize online thermal monitoring of industrial continuous casting molds and 2) to create a new research tool to investigate meniscus behavior so that defect formation can be better understood.

A preliminary design for the sensor and its installation into a continuous
casting mold is proposed in Figure 1. The sensor pad containing the wire junction points will protrude out of the top of the copper mold and nickel plating, thereby making wire connections possible after plating. These wires will be run a short distance to a wireless transmitter which will transmit the signal to the control-room operator display or other computer. A wireless thermocouple system that is compatible with the sensor was obtained from MicroStrain. The system supports simultaneous data transmission from multiple sensors, theoretically allowing thousands of sensors to be installed into a mold.

Return to Top

Experimental Studies: sensor strip attachment and electroplating embedding

Before any research can be carried out to understand the behavior of embedded sensors, it is imperative that a robust method be developed to embed sensors in manufacturing tooling. Therefore, there is an absolute need to refine the embedding process through plating studies, as the sensor strip will be attached to the copper mold via the electroplating process used to apply the nickel coating layer. Any attachment method must provide a secure bond between the sensor strip and the copper mold face, have no air gaps, survive the acid pretreatment steps, and allow the sensor strip to be plated successfully before the copper mold can be put into service. It is possible to attempt to plate the sensor strip, which is fabricated encapsulated by nickel, by placing it on the copper mold face and submerging the mold face into the plating tank of the mold manufacturer. Two different trials spanning a range of aspect ratios were performed to evaluate the plating ability to attach a nickel strip to a copper substrate via commercial nickel electroplating. It was proposed to repeat the initial plating study with aspect ratios down to 0.33 in order to observe complete filling. The analysis of this plating study is currently being conducted.

Another possible method is to attach the sensor to the mold face before the plating operation is carried out. An electrically and thermally conductive silver paste (k=109 W/m-K) used in adhesive and coating applications to 1200°F has also been considered due to the ease at which trials can be conducted. Ultrasonic welding and diffusion bonding have been proposed as possible solutions. An ultrasonic welding equipment manufacturer agreed to perform an ultrasonic welding experiment to determine the feasibility of attaching a nickel strip to a copper substrate. Thus far ultrasonic welding has not proven itself an acceptable solution to the sensor attachment problem.

Fiber optic sensors are being pursued as an alternative to sensor strips. As with sensor strips, embedded fiber optic sensors have the advantage of real-time monitoring at critical locations as well as immunity to electromagnetic interference and resistance to hostile environments. Initial attempts to embed optical fibers have been met with success. Further trials with fully functioning fiber optic sensors have been planned.

Once it has been shown that a sensor can be successfully plated within the nickel layer over a copper mold, the bond between the sensor, nickel plating layer, and copper substrate can be tested. An initial check of the sensor will be performed by the mold manufacturer by immersing the copper substrate in water and applying a flame to the nickel plating layer. Any deterioration of the plating layer such as spalling or cracking will indicate a failed plating attempt.

Return to Top

Modeling Studies: Consequences of Gap Formation and Initial Solidification Phenomena

An incorrect plating procedure can result in an air gap forming between the sensor strip and copper substrate. In service an air gap present in the continuous casting mold can limit heat transfer and cause a localized high temperature region near the sensor. Such a temperature increase can contribute to plating layer failure or lead to the sensor spalling off. Computational methods have been utilized to quantify this behavior. This is needed in order to evaluate the maximum size of gap that would still allow performance of the sensor.

The final task will be to interpret the signals from the new sensor. One benefit will be to discover new insights into meniscus phenomena and defect formation, which arise at the meniscus. In the meantime, work to improve understanding of meniscus phenomena is proceeding at the University of Illinois using computational models, metallography, and microscope analysis of plant samples, previous laboratory experiments, and conventional temperature measurements. A new mechanism for the formation of oscillation marks has been developed, which involves freezing of the meniscus. This work is described elsewhere [1-4].

Eventually, sensor signals obtained online at the commercial caster will be correlated with defects in cast steel found by evaluation of the cast product. This work will have the most commercial impact. Efforts will be made to identify characteristic signals or 'signatures' of the formation of particular individual defects. This will advance the technology towards the ultimate commercialization of an 'expert mold'. Improving this important process even slightly has a huge potential impact in energy savings, yield savings, steel quality, and efficiency improvement, because this particular process is used to produce several hundred million tonnes of steel every year.

Return to Top

Overview of Findings

Experimental Studies: sensor strip attachment and electroplating embedding
• The wireless system was tested in the presence of an electromagnetic field at a commercial steel company and was shown to maintain communication between the thermocouple node and base station transceiver.
• During the initial plating study, nickel adhered well to the top, bottom, and sides of the nickel strip, as well as the copper substrate, as depicted in Figure 2.
• Often, when electroplating onto a thin strip suspended a small distance from the substrate, a void commonly forms in the gap, due to the starvation of nickel atoms when two sections of growing grains impinge just past the edge of the nickel strip. This also causes the void width to extend wider than the strip width, seen as seams at the edge of the void.
• The rate of plating is greatest in areas where sharp corners exist, due to higher current density.
• More plating reaches under the nickel strip as the sensor width decreases and/or gap thickness increases. More specifically, more plating reaches under the nickel strip as the aspect ratio, defined as the strip width over gap thickness, decreases.
• Complete filling under the nickel strip is predicted to occur when the aspect ratio is less than or equal to one.
• A sensor strip can be attached to a copper substrate via conductive silver paste and successfully plated over without any air gaps. However, the time and skill involved in this process is considerable.
• According to the ultrasonic welding equipment manufacturer, past experiences have indicated that ultrasonic welding near, around, or on top of sensors has lead to irreversible damage.
• Although the ultrasonic welding equipment is able to attach the two dissimilar metals, the machining pattern caused by the weld horn is unavoidable.
• Due to their non-conductivity and geometry, fiber optic sensors can be plated over without an air gap developing (Figure 3).

Modeling Studies: Consequences of Gap Formation and Initial Solidification Phenomena
• It was found that an air gap in the nickel plating layer can cause stress to increase by 19%.
• A wider gap makes it more difficult for heat to conduct around the gap, increasing the temperature at the hotface: doubling the width of the air gap increases the hotface temperature by 65°C, while doubling the thickness of the air gap increases the hotface temperature by only 5°C.
• Oscillation marks and hooks which comprise the initial solidification structure form due to meniscus freezing and overflow. This brings heat to the meniscus in a characteristic periodic heat flux, which is usually increasing during the negative strip time. A detailed computational model of this has revealed the fundamentals of this behavior and a detailed mechanism for the phenomena. This is summarized in 4 publications [1-4]

Return to Top

Training and Development

Through working on this project, several researchers have been educated in the development and application of computational models to solve practical engineering problems. These include PostDoctoral Research associates, PhD students, MS students, and undergraduates.

Specifically, these include Michael Okelman (MS student), who worked exclusively on this project and Joydeep Sengupta (PostDoc), Claudio Ojeda (PhD student), and Gogi Lee (PhD student), who worked part-time on this project. Two of these (JS and CO) are now working as researchers in the industry, where they are putting their knowledge from this project to good use.

Return to Top

Outreach Activities

The results of this project were presented at the annual meetings of the Continuous Casting Consortium to the steel-industry members and at conferences, giving the student presenters exposure to industry, while simultaneously disseminating the project findings. In addition, some of the results from this work (those on initial solidification mechanisms) have been incorporated into the short course to the steel industry (Brimacombe Continuous Casting Course, in Vancouver, Canada) which is presented annually to industry attendees by a team of 5 instructors (which includes the PI of this project). Attendance during the first 2 years of this project consistently exceeded 150 people, and ranges from operators who implement technology on the shop floor to company researchers who work on improving that technology. The audiences have been mainly from steel companies in the U.S.

Return to Top

NSF Publications:

Li, X., C.W. Wong, D. Dornfeld, and B.G. Thomas, “Research on Subwavelength Microphotonic Sensors for In-situ Monitoring with High Spatial and Temporal Resolution in Manufacturing Environments”, Contacting and Solidification in Casting-by-Design”, Proceedings of 2006 NSF Design, Service, and Manufacturing Grantees and Research Conference, St. Louis, Missouri, July 24-27, 2006, 9p. Click here for a PDF version. (644 KB)

Return to Top

Journal Publications:

Sengupta, J; Thomas, BG; Shin, HJ; Lee, GG; Kim, SH, "A new mechanism of hook formation during continuous casting of ultra-low-carbon steel slabs", METALLURGICAL AND MATERIALS TRANSACTIONS A-PHYSICAL METALLURGY AND MATERIALS SCIENCE, vol. 37A, (2006), p. 1597.

Return to Top

Book(s) or other one-time publication(s):

Ojeda, C., J. Sengupta, B.G. Thomas, J. Barco, J.L. Arana, "Mathematical Modeling of Thermal-Fluid Flow in the Meniscus Region During an Oscillation Cycle" , bibl. May 1-4, Cleveland, OH), AIST, Warrendale, PA, Vol. 1, pp. 1017-1028., (2006). Conference Proceedings article Published of Collection: , "AISTech 2006 Steelmaking Conference Proc."

Sengupta, J., B.G. Thomas, H.J. Shin S.H. Kim, "Mechanism of Hook Formation in Ultra-low Carbon Steels based on Microscopy Analysis and Thermal-stress Modeling" , bibl. Vol. 1, AIST, Warrendale, PA, (May 1-4, Cleveland, OH), 2006, 903-914., (2006). Conference Proceedings Article Published of Collection: , "AISTech 2006 Steelmaking Conference Proc."

Thomas, B.G., J. Sengupta, C. Ojeda, "Mechanism of Hook and Oscillation Mark Formation In Ultra- Low Carbon Steel" , bibl. Baosteel, Shanghai, PRC: May 25-26, 2006, Shanghai, PRC. p. 112-117., (2006). Conference Proceedings Article Published of Collection: , "Second Baosteel Biennial Conference"

Return to Top

Return to Top