In USA:  Morgan  |  Roll pass software  |  Cascade  |  .Net & Metal Pass  |  Consulting  |  Road ahead

Move to USA. I went to Morgan Construction Company for interview shortly before my Ph.D. defense. Morgan posted an advertisement in a German magazine "Stahl und Eisen" to recruit a high profiled engineer worldwide. Before this visit, the Vice President for Human Resource kept phone contact with me every day (3:00pm in Germany, 9:00am in USA) to make sure my arrangement to visit USA was OK. At that time Danieli was asking me to arrange Visa for another visit to Butterio, Italy to sign work contract, and Dr. Gartinger from Mannesmann Demag also asked me to join his team. Prof Schwenzfeier, director of Institute of Plastic Deformation and Plant Machinery, University of Leoben, had sent me three letters and invited me to work as a Post-Doctor researcher in his institute. Invitations I received also included those from Ohio State University, and a US company who wanted me to do the Metal Forming research, and so on. I thank so many colleagues and industrial partners who showed interest in working with me. Among the countries, USA is the place I prefer to work in. At that time, the general job market in our industry was still bad. A job posting would often led to over 1000 applicants.

The three day's interview in Morgan was under tight schedule. In the first round I met over 20 people, one by one following a check list. Those 20 people consists of all the VPs and Top managers, including the senior VP Daniel Morgan, one of the Morgan brothers who own Morgan. With a satisfactory result, I flew to Kansas City to meet the senior vice president Mike Shore, who was in charge of all the technical aspects of the company. Mike and I talked for six hours (8am - 2pm) on various technical topics. Mike told me the reason to create this position was that he would retire in seven years, so he wanted to have a young engineer to train. He said at the time Morgan didn't have anyone strong enough to continue what he was doing. At the end of the talk he declared he had already decided to hire me, and he wanted the hiring formalities to be finished before I left for Germany. My return trip to Germany was re-arranged to allow Morgan one more day to finish the hiring processes (including offer letter). In the last day Morgan particularly arranged a hotel room in downtown Boston (50 miles away from the company) for me to have a look at USA. I was impressed with the pay and a sign bonus of around $10,000. I deeply thank Morgan brothers, president Philip Morgan and the senior VP Daniel Morgan, who treated me just like their friend. Whenever Philip met me, he would stop to talk to me. Once he heard somewhere that I moved from Worcester to Shrewsbury, so he stopped to ask where in Shrewsbury, and commented "it is almost country". During lunch he once bypassed my manager's manager and stopped to talk to me. On my personal trip to China, Morgan hired a transportation company to send me to airport and, at my return, picked me up from the airport.  From Worcester (where Morgan's headquarter is) to Boston airport it is over one hour's distance.

Work for Morgan. Morgan realized that there were some problems in its rolling mill models and roll pass design procedures. So I was asked to fully examine its rolling process models and roll pass design procedures, and to do improvement. The top managers in Morgan paid great attention not only to my own mill modeling experience, but also the materials I took from Germany. My former institute is specialized in the steel rolling mill research. It built a lab mill in early 1980s, a four-stand high-speed continuous mill that can roll up to 70m/s and is with controlled cooling capability. Even today it is probably still the most sophisticated test mill in Germany. By the time I joined Morgan, the institute had spent over 15 years to do the research with the mill. The research was led by the institute director, Prof. Dr. Arno Hensel (my academic adviser). Prof. Hensel, who had worked in that capacity since the age of 39, is one of the most successful researchers on rolling mills, in Germany and the world. As I came to work for Morgan, I took nearly 10 boxes of technical documents I collected from Germany. This also includes some unpublished writing from Prof. Hensel with the condition that I can use the results but should not publish them. For wire rod rolling alone, for example, I took over 40 Ph.D. theses from Germany with me. I, with my good training and with the research results from Germany, very quickly found the problems.

Typical problem was that in the high speed rolling, traditional calculation procedures for low speed rolling, both for force prediction and metal flow estimation, were not valid any more. Traditional flow stress model is only valid up to 100/s strain rate, while the high speed rolling may have a strain rate over 3000/s! Many passes, such as those in Reducing/Sizing Mills (RSM), it is almost Round-Round pass sequence, and no traditional model can fit. Using old model would lead to 30% over-capacity of the mill, which means the cost for a customer to build a mill could be tremendously higher (even over 50%) than needed. Also, in the high speed rolling, the low friction and high heat generation cause the deformation pattern (both width spread and forward slip) totally different from traditional rolling. Inaccurate prediction of the rolled cross-section and rolling speed may lead to the wire between the stand to be loose and this causes the flutter (mill vibration). A high speed train (No-Twist Mill), with 4-12 stands, is driven by a single motor, and so the problem cannot be solved by adjusting the speed of each individual stand.

Unfortunately, my results and good intention to help Morgan, were probably a part of the reasons that caused my direct supervisor, the Morgan's chief engineer since 20 years, to be removed. The former chief engineer and the director of the roll pass department, had actually pretty good experience in steel mill, though, like every US engineer on steel rolling, his theoretical understanding on the rolling was limited. My understanding on steel rolling was derived from systematical training with my B.S., M.S. and Ph.D. all on steel rolling. The over 30 classes on steel rolling I completed fully covered the metallurgical fundamentals (2D and 3D dislocations, grain size, microstructure, recrystallization, grain growth, etc.), mechanical fundamentals (flow stress, temperature, friction, spread, forward slip, force, etc.) and rolling technologies (e.g., plate/strip rolling, rod/shape rolling, and pipe/tube rolling, mill equipment and design, and hot/cold rolling). In particular, during my 3D FEM simulation and experimental verification of the rolling process, I closely observed the metal deformation in the roll gap, day in day out; there was hardly anything I hadn't observed. That led to my development of the simplified model, even though for quite a while, only several rolling mill experts such as Kopp and Hensel understood the model, while the others simply had hard time to get it. In comparison, most US engineers, which were usually general mechanical or metallurgical engineers, had never a single class on rolling (some schools may provide several hours of introduction of rolling process).

We soon got a new director for the roll pass department. He was initially an engineer for mill guides and was appointed as the Roll Pass chief after three months of basic roll pass training in Japan. He told me he would prefer a simple way for the parameter calculation and for the roll pass design. I demonstrated the high accuracy of my model with the experiment data, mostly, the field data we collected from the field, such as then ASW in Cleveland. I understand that many people, due to the limited training on the rolling process, overly simplify the process and prefer the easy solution. This also gave me the idea that I should develop roll pass software to handle the complicated rolling process problems and to allow the users to do the design in a easy way.
 

In USA:  Morgan  |  Roll pass software  |  Cascade  |  .Net & Metal Pass  |  Consulting  |  Road ahead

Mathematical Modeling. The major responsibility for me in Morgan was to design process models to improve it's design procedure, so I generated an entire set of rolling process models to serve the roll design aimed at making the designs more reliable. This set of the models serve as the basis for my later model-based software development up to today. The models are under inseparable combination of mechanical, metallurgical and rolling process aspects. Both empirical models and numerical approaches (with finite element method) are used. The models include those to predict metal flow (spread for shape change, forward slip for speed determination), force, torque and power, as well as temperature and microstructure formation (recrystallization, grain growth, etc.). Test data used for that development were e.g. Morgan lab mill data (from a five-year rolling test) and results of a 15-year study for rod and bar rolling in my former institute in Germany. Extensive metal forming knowledge was involved in the process models, such as tension and temperature corrections for spread and forward slip, essence of flow stress and microstructure, as well as the shape factors for load and torque prediction. Such models may be beyond the reach of a normal mechanical engineer. For details of the models, see "Process Modeling" part.

Usually, the people use different set of models to calculate roughing train, intermediate train and finish train. My models, under consideration of the speed, etc., fit all three trains. The models have been verified with the No-Twist Mill (NTM) data collected from former ASW (Cleveland, OH). A research engineer spent several months in ASW to measure the rolled cross-section in the NTM pass by pass, and I also spent quite a while in ASW to collect and process data. I was sent to ASW primarily because the other colleague had the problem to process data to find a model. The prediction of the cross-section with my model excellently match the measurement.

FEM Analysis. I applied the FEM model established during my Ph.D. study to Morgan's projects. This was to investigate the capability of FEM in rod rolling simulation. Local metal flow, geometry, load, power, temperature and microstructure, etc. were predicted. Results were verified with earlier measurement, e.g., Morgan lab mill data. During the microstructure simulation, microstructure models were coupled with the FEM model.

Computer Training. In the summer 1998, Asia economic crisis heavily hit Morgan's business and many people were leaving Morgan voluntarily. For a while we had seeing-off lunch every day. In fact, right after I came to USA I predicted that US steel industry would be in trouble because I had an impression that US didn't work hard to develop steel technology at all. In addition, I didn't see the people who understand rolling process except one person, the Morgan's senior VP Mike Shore, whom I talked with for six hours during my interview. At the time I had decided to do some computer training. For one reason I wanted to utilize the rolling process models I had developed because some colleagues seemed to feel the model was a bit too complicated; for another reason I needed to prepare for my job security before the trouble would come to the steel industry. Sure enough, in 2001, due to the poor economic situation and lack of competitiveness, 75% of the largest US steel company went bankrupt.

I took computer classes in the evening and also arrange time to develop roll pass software. I also passed around my resumes to see any company would like to hire me to write mill software. Top steel companies, such as US Steel, LTV, Nucor, etc., gave me the job interview, but no one asked me to do software programming. In talking with Nucor, I demanded that if Nucor would pay me $100K per year, I would work for it (and would give up the computer training). My manager-to-be in Nucor seemed OK to my pay request, but his manager didn't agree, though they both would like to hire me. LTV technical center asked me to do FEM simulation, but I only had interest to write (roll pass) software. USS Technical Center had a project to investigate the cause of the surface problem of the rolled steel, and I had no interest, either.

My workload was extremely heavy because I took too many classes. Interesting is, the tuition in this year could be fully covered by a US tax policy HOPE (up to $2000 per year, available for every tax-payer). After seven classes were finished, I decided to look for an IT job so that I could do the training by doing the job. Surprisingly, I got a job offer within 10 days from the start of job search, and the pay was based on Ph.D. with three years of experience! That's was Cascade Steel Rolling Mills (CSRM). I responded to a job posting for a steel plate mill in Oregon (it should be OSM), but for whatever reason, the recruiter switched the job provider for me, so during the phone interview I was pretty confused. It was so easy for me to switch to be a software engineer. Maybe because the IT industry was booming, and maybe people considered my five-year experience in a computer laboratory on rolling process simulation (during my Ph.D. study).

Roll Pass Software Development. In today's technology level, rolling speed has been over 100 m/s (20,000 ft/min). Traditional way of roll design based on rough estimation is not sufficient any more. Many important influence factors have to be taken into account in roll design. Otherwise, various problems, such as flutter (mill vibration), would occur. This happened in many steel companies, such as former Inland Steel and American Steel and Wire (ASW). On other hand, while considering various influence factors, software has to be generated, because the influences are very complicated. Based on this consideration, I generated a design program (initially based on Microsoft Excel) with the metal flow models I developed. Metal deformation was calculated pass by pass. Pass overlay was drawn automatically. In tied stands, tension between two stands was also graphically presented. The program was generated during several roll design jobs of mine. Design projects were performed with it. One example of the projects is my roll pass design in Morgan for former GST 19" bar mill in Kansas City. It's an 8-stand-mill with the last four stands tied (driven by a single motor). It was required to roll 28 finish sizes, including rounds and squares, from only 3 entry stock sizes. As an existing mill, gear ratio among stands could not be changed. Therefore tension between stands was not avoidable and the tension correction of metal flow had to be taken into account to reach the required accuracy. The average reduction was so high with the billets available, that the reduction was near the limit in many passes. In such a complicated case, it is unimaginable to perform accurate design, if there is no design software available with sufficient process models in it.
 

In USA:  Morgan  |  Roll pass software  |  Cascade  |  .Net & Metal Pass  |  Consulting  |  Road ahead

Work as Software Engineer. I accepted a job offer as software Engineer from Cascade Steel Rolling Mills (CSRM) to implement Steel Mill Level 2 system. Before this, I already completed lots of evening classes on VB/C/C++/MFC, etc. This position in CSRM was a pretty good one for me because I prefer to work in my industry (Metal/Steel related) as a software engineer.

Steel Mill Level 2 Project. The CSRM's Level 2 project involves EAF, LMF and Caster manufacturing stages. It collects data from numerous PLCs and Chemical Lab, populates data in Man-Machine-Interface (MMI) operators computers to assist operators to run steel manufacture. Scheduling, Steelmaking recipes, chemical analysis, personnel, as well as energy and material consumptions are also the key functions. The MMI applications (Client) were implemented with Visual Basic, while the Level 2 server applications were written with Visual C++ and MFC. The back-end database server was Oracle. After Level 2 was put into operation, there were lots of work for debug (until middle 2002), Oracle database data quality management and data populating to the business systems. I decided to implement this part of applications with Microsoft .Net. In the mean time I was assigned to design the rolling mill Level 2/Level 3 system. I was the one, the only one that fully responsible for the company's Level 2 automation system; besides Level 2 and shop floor computer support, I was the only software programmer, and the web developer and web Master. The rest colleagues in the IT department were: an IT manager, a system supporter, an Oracle DBA and a hardware technician. For all shop computer, the technician only put the hardware pieces together and I would do the rest. My work was recognized as with higher quality, and soon, all the mission critical applications/systems were assigned to me. It was my responsibility to add middleware (e.g. PLC card), to configure resources (to make sure the PLC card didn't have resource conflict with other hardware), to install and configure Windows operating system, and then to install touch screen drivers and to configure touch screen. After all those were done, I started to install Level 1 and Level 2 software. After those steps, there were another 10 different configurations, such as to set up auto logon, to specify the Level 2 servers and database sever, and so on. I assumed that the operators' computer skill level was zero; if anything went wrong, they only needed to reboot the HMI computer and so everything should come up automatically. I made disk clone for every shop computer; so if any computer was dead, I only needed several minutes to rebuild it. I carried company's pager with me. My computer setup was done in this way: for any of the common problems, the computer would send a message to the Outlook server and the Outlook server would call my pager, so I would actually receive a phone call. Based on the phone number from which I was called, I would roughly know what was the problem. Frequently, the Oracle database server lost connection with one of the Level 2 servers and this troubled me for quite a while, until we made big design change to make sure the Level 2 server would keep the SQL scripts for inserting/updating data for up to several days. As long as the system regained the connection, the Level 2 server would automatically re-run the SQL scripts to send data into the database. The DBA could also manually run the SQL scripts to send the data. This part of the improvement counted for quite a portion of my Level 2 development. If one never worked on the Level 2, he wouldn't realize that this was actually a big part of the work.

Web-Based Applications and Related Training. I also developed many web-base applications for CSRM, with Microsoft .Net. I started using .Net as it was still in the beta 1 testing stage. Examples of the web-based applications include: (1) an application to manage and populate Excel spread sheet onto the intranet web site by reading an excel file as a database; (2) Oracle data management tool with which engineers and operators can read data; if modification occurs it can modify the data and move the original data to the history tables; (3) an application to manage engineering drawing files and display all drawings (in AutoCAD format) on the web screen; (4) a tool to populate PLC data directly onto web screens. Besides those web-based applications, I was the sole web developer and web-master responsible for both the internet and intranet websites, which were designed/developed by me.

Cascade Steel sent me out to do some necessary trainings. For example, I attended a three-day training for Windows 2000 administration in a training company called New Horizons. My training in New Horizons also included a class on networking fundamental. The rest classes I attended were mainly arranged by myself, some of which were paid by Cascade Steel. In fact, most part of the work I did, especially that in the first half of my work years, was a great opportunity for learning.

Though my initial motive for software engineering was to utilize the process models I developed, during my work on the Level 2 supervisory system I realized, that besides the rolling process modeling, the data collection and data communication would also be a very prosperous area to work in. In the mean time US economy moved into recession, so I intentionally delayed to develop roll pass software and decided to establish the Level 2 development as the second work area of my future career. In the late 2001 I expressed my intent to my manager that I would consider to leave the company, but he managed to persuade me to stay. Though I believed that the economy would be recovered in 2002, at that moment it was still slow. In the last 3 months of my work for the Cascade Steel, I actually moved from McMinnville region (where Cascade Steel located) to North Portland, one and an half hours away. Most computer classes were offered in the Portland region. Before the move I needed to drive over an hour to attend a computer class. During this three-year employment I collected and processed rich resources on the steel mill Level 1, Level 2 and Level 3, in combining with the Level 2 project I worked on, and added with the rolling mill process models I developed years before, I wrote my second book - Steel Mill: Process Modeling and Computer Application. With such hot topics and my first-hand materials, I was confident the sale would be great. However, I eventually decided not to publish this book, at least in a short term. I didn't want to give away so many valuable materials and ideas for free, but planned to market the results in some other ways. In those years I have kept adding new materials to it, and most recently, the flat rolling shape control models and plate/strip mill Level 2 technologies.
 

In USA:  Morgan  |  Roll pass software  |  Cascade  |  .Net & Metal Pass  |  Consulting  |  Road ahead

.Net training. My computer experience started from Unix and Fortran. My computer training before my work as a software engineer was focused on the Microsoft Foundation Class (MFC) for Windows application development. My development for Cascade Steel was for Windows application using MFC. Since a portion of my work was also web application development, and Java with accompanied platform was the primary tool for the web development, so I initially selected Java as development language for web application. However, right after my two Java classes (one in the undergraduate Level and another in the graduate Level) were finished, Microsoft was going to release its primary development platform, the .Net. Though ASP was weaker than JSP, the ASP.Net seemed much more powerful than JSP. So I started to study .Net from its alpha test version and wrote applications with the Belta version of the .Net.

I attended one of the earliest .Net training classes in Portland region. It was taught by Dr. Arnie Roland, who holds his Ph.D. on computer science since 20 years and earned quite a few Microsoft certifications. Arnie was an industry consultants who helped Microsoft in many development projects. In this training program, he was hired by Portland Community College (PCC) in its Open-Campus training program, which targeted software engineers rather than registered PCC students. The PCC's Open-Campus training was one of the primary training program for .Net in the Portland region. From that time on, I took every one of all its training classes on .Net and related, including the classes for: (1) Microsoft Certified Solution/Application Developer (MCSD/MCAD) for Web Application; (2) MCSD/MCAD for Windows Application; and (3) Microsoft Certified Database Administrator (MCDBA). Courses were concentrated on the .Net and SQL Server related technologies, such as ASP .Net, VB .Net, XML Web Service, COM+ Server Components, SQL server programming and administration, etc. For the purpose of developing a website, I also attended several classes such as HTML programming, Shopping cart programming and  Web design with Adobe PhotoShop (for graphic design).  See Info Tech - Courses Attended pages. The classes were all held in the evening, from 5:30pm to 9:30pm, because students in the class had their job during day time. Certification training was one of the primary ways for a software engineering to move up to the senior software engineer and system architecture. I usually attended classes for 2-3 times every week. In the class I was the one who asked lots of questions, so I believe every instructor would remember me for a long time. Though all students in the class were experienced software engineers, some of them in quite senior positions, I still seemed to be one of the most knowledgeable persons, partially due to my longer .Net involvement history and partially because I had taken much more classes than any of them. The training materials (books, lab projects, etc.) were all from Microsoft, and the instructors held at least one certification from Microsoft, some 3 to 4 certifications (like Dr. Arnie Roland).

Development of metalpass.com. Development on the metalpass.com traced back to the very early years. In 1999, I registered my first domain name metaldata.com, intending to post information on my rolling process models and on the metal software I would develop based on my rolling process models. Later, I was amazed by the Microsoft Technet, which provides support information on the Microsoft products. I tried to follow the idea, and so I started to develop metal technet (metaltechnet.com). Then I realized it would also be nice to add business information onto the website, such as the metal directory. So since 2003, I have been using metalpass.com to post both technical and business resources. In the internet world, metalpass.com site was considered by many tracking agents to have been started in 2003. In fact, its history is much longer. I believe that metalpass.com is the one of the largest commercial websites in the metal industry. Metal directory alone, has over 50,000 records (still under expansion, for example, with 1000 steel plants as the recent addition). Metalpass.com also has exciting web-based design software running on it, such as the roll pass design programs and a finite differential program to determine temperature evolution during hot rolling and controlled water cooling. If you are an IT engineer, you'll be fascinated with the the software technologies it involves. The development of the metalpass.com were performed parallel with my intensive .Net training - this was also a great help for my .Net training.

A great portion of time were actually spent for web interface design and the data processing. I also wrote web/window-based programs and tools for data processing and process modeling, for example, a tool to measure flow curve coordinates and to send data into a database, and a program to model flow stress data into various form of formulae depending on temperature, strain and strain rate. By August 2004, I have finished my computer training (most classes were completed by end 2003). I planned to finish the metalpass.com development at the end of 2004, but I actually reached the most of the current web functions in the Feb. 2005. After several months of website improvement, I started to develop web-based roll pass software in the middle of September 2005 and finished the major applications at the end of the June 2006. Those applications running on the metalpass.com are a portion of the software programs I designed in that period.
 

In USA:  Morgan  |  Roll pass software  |  Cascade  |  .Net & Metal Pass  |  Consulting  |  Road ahead

Level 2 consulting for OSM. Oregon Steel Mills, Portland, OR, runs its plate processing facilities, with its Level 2 system supplied by Tippins followed by OSM's own continuous improvements. It is a sophisticated Level 2 system with which, the production, with an annual sales of about $800 million, is running by only about six operators. The source code for this Level 2 system has over one million lines. Such a system may need nearly two million dollars to purchase in today's cost level. The system has some problems when the rolling temperature is relatively low, and in the hard and thin grades. I spent about four months to examine its source code and to investigate problems via log files and databases. Since the database usually has over a million records for a year, so I wrote various programs using .Net to process the data.

Three major problems were identified: logical problem in its adaptive learning, metallurgical problem in the low temperature rolling (including holding), and flow stress valid range problem. The Level 2 system used adaptive learning to recalculate flow stress coefficients, namely material factor C1, temperature factor C2, strain factor C3 (or also C5), and strain rate factor C4. Two issues were involved in this aspect. One was that if a factor (such as C3) was not used for learning, a zero value was used instead of a non-zero number. This pushed the contribution of this given factor (strain influence, for the C3) to others and caused fluctuation of the other factor (e.g. C4).  Another issue was the dependence of C3 and C4 on each other: either higher strain coefficient (C3) plus lower strain rate coefficient (C4), or lower C3 plus higher C4, may work. Those two issues limited the quality of the learning. The metallurgical problems were also in two areas: in the low temperature rolling where an incompleted recrystallization caused a retained strain to occur, similar to the partial work hardening. The second problem was caused by the metallurgical change during the period when the steel was held in the air between passes.

The first stage of the solution to the above-mentioned problem was: I designed a guided two-parameter learning procedure, with optimal values for C3 and C4, and used C1 and C2 as the learning factor. More specifically, I designed 6000 sets of flow stress coefficients C1, C2, C3 and C4, for the 2000 model grades and for three temperature zones in each model grade. The results achieved after this stage were: 80% passes were within 5% force error, 90% passes below 10%, and 99% passes below 15% error. This has been much better than those before my improvement: 57% passes within 5% error, 87% passes below 10%, and 94% passes below 15% error. In particular, this modification eliminated the problem that in the hard and thin grades, force error sometimes reached 40% and thus the quality of the rolled steel was very bad. There were still 1% passes with over 15% error. This was primarily due to the metallurgical evolution during the hold, and due to the flow stress formula valid range problem (strain below 0.1 or over 0.5). My solutions to those problems were to modify the force factors for the hold process, and to extend the valid range of the flow stress formula by modifying the strain. This further increased the accuracy of the Level 2 model prediction. See the related reports for OSM, in www.metalpass.com/consulting.

Besides those improvements, I also provided a long lists of suggestions for the Level 2 system architecture, and database and data structures design, as well as other model improvements such as temperature, shape factor and roll flattening. In addition, I proposed projects and corresponding work plans for Camber reduction, rolled steel property prediction, and heat treating furnace software development, and so on. OSM has set up a team to deal with the camber problem, with measuring cameras, which I proposed in a writing on Camber.
 

In USA:  Morgan  |  Roll pass software  |  Cascade  |  .Net & Metal Pass  |  Consulting  |  Road ahead

Back to the East. In the summer of 2007, I finally moved back to the East, to the manufacture concentrated region. I have moved here in Pittsburgh, the Heart of Steel. In my eight years of stay in Oregon I focused on the software engineering improvement. The region I lived was very computer-centered. Intel was right the neighbor, and Microsoft was within the driving distance. There were also IBM and dozens of electronic companies formed by people who left Intel. Dozens of friends of mine, and the people closed to me, work for Intel, IBM and Microsoft, etc. In the computer gathering we frequently received guests from Microsoft who promoted the .Net. I moved to Oregon as an entry level software engineer, but now, after eight years, my training cycle has been finished. In the old Chinese action stories, a person suddenly disappeared from the community - he was hiding in the deep mountain to improve his fighting skill. Years later he was back to the world and his fighting skill became unequaled. I was also hiding in Oregon for eight years. During the day I was a software engineer, and at night I was in the computer lab for IT training. I didn't miss any single one of the Microsoft certification training classes on .Net for Web-based applications, on .Net for Windows-based applications, and on database administration, as well as many computer system classes. Though my classmates were all experienced software engineers, I was still the one that gave the instructor deep impression, for my questions and my skills. My first station in the USA was in the East. Now I am back. My 35 years of training, especially the work and training I received in the past 25 years on the steel rolling mill processes, mill process modeling and computer programming, have provided me with sufficient capacity for my development on steel mill software system.

In my trip to Pittsburgh, I stopped for four days in Colorado School of Mines to collect rolling mill related resources. Thanks for the colleagues there to be very friendly and supportive, I got useful books, related Ph.D. theses and valuable data from CSM's computer systems. After I arrived at Pittsburgh, I quickly collected useful materials from Pittsburgh's three primary libraries. In the past months after I came to Pittsburgh, I processed over 10,000 pages of documents related to the steel mill modeling and steel mill automation. This greatly increased my knowledge and resources. The Basic Metal Processing Research Institute (BAMPRI) of University of Pittsburgh, consisting of three processors and a team of research assistants, has joined the hand with me to perform steel rolling process modeling. American Sensor Cooperation (ASC) also has cooperation with me in the automation and measuring techniques.

Road ahead. My initial motive for the computer training was to develop roll pass program. During my work for Cascade Steel Level 2 project I entered the world of Level 2. Considering the fact that in recent years the business for wire rod rolling was slow, while the flat rolling, such as strip and plate rolling, was very profitable, so I intentionally expanded my initial work area of wire rod rolling and shape rolling to include the flat rolling. Typically, flat rolling is much easier than the shape rolling, though it also has its own technical difficulties. Metal deformation and force prediction of the flat rolling process are easier than that of the shape rolling. Technical fundamentals such as microstructural evolution and flow stress determination are exactly the same for the both processes. One of the technical difficulties for the flat rolling is that related to the roll deformation, such as crown control (roll crown and roll bending, etc.) and thus the flatness and straightness of rolled strip/plate. The metallurgical issues (including the controlled cooling) for both flat rolling and shape rolling are also similar. The roll pass program for flat rolling (typically, the Level 2 model) has more factors tied together than that for wire rod rolling. FEM is one of the most effective ways to solve the problems such as roll deflection and roll flatening, and FEM modeling is one of my technical strengths which I worked on very successfully for the DFG (Germany) during my Ph.D. research. Since I was moving from the harder part of the rolling process (shape rolling) to the easier one (flat rolling),  technical difficulty is not my concern.

Though my primary focus in the last eight years was on software engineering, I have never left the steel industry. I was always working on computer application for the steel mills. Up to now I have completed nearly 100 steel mill and related projects (more projects and descriptions are to be uploaded). Considering my steel mill experience and computer training, my past links with China, Germany and USA, especially with colleagues in both engineering companies (SMS-Demag, Danieli, Morgan, etc.) and steel plants, and my past work for DFG (Germany Research Association) and recent involvement for NSF (National Science Foundation, USA), I believe my learning period is completed. Now it is time for me to help my industry.
 

In USA:  Morgan  |  Roll pass software  |  Cascade  |  .Net & Metal Pass  |  Consulting  |  Road ahead

More reading for "In USA"