Richard Skeirik 
Career Narrative

My career evolution began as early as high school, where I excelled at math and science, especially chemistry. A National Merit Scholar, I chose biochemistry as my undergrad major, thinking I would work in life sciences. But my experiences as an assistant in a psychopharmacology lab convinced me not, and after my BS I switched to chemical engineering and entered a graduate program. My taste for exams and committees at that time was limited and I opted for a masters degree with a thesis aiming for an industrial career. My masters research explored the dynamics of chemical reactors, using dynamic simulation to search for selectivity enhancement in periodically forced CSTRs with complex, nonlinear kinetics(1). I also developed a computation simplification for numerical simulation of high molecular weight polymerizations(2), which years later I applied to DuPont's Terathane process.

1982: Catching a last branch as the job market tumbled off a cliff entering the 1982 recession, I went to DuPont's Experimental Station research center to begin a 22 year career endeavor.

My first two years I worked to develop potential new processes for adipic acid production. Working with organo-metallic chemists, I helped to shape processes based on their chemistry. I designed and built a small pilot unit for one synthesis option(3); later I laid out flowsheets, built mainframe-based sequential modular flowsheet simulations, and then costed and evaluated profitability of building and operating these processes(4). In parallel, I computerized instrument data acquisition in the laboratories, implemented minicomputer control of a major pilot plant, and installed the first personal computers and workstations.

1984: I moved to DuPont's Old Hickory plant where I was assigned to a world-scale production unit for the monomer used in polyester. Already running mini-computer based controls, including a form of internal model control, the plant faced continued pressure to both improve control and meet ISO standards. Online analyzers had already be problematic; I integrated my dynamic reactor modeling with the plant's historical data acquisition system, and built a real-time state estimator for catalyst composition(5). We closed control loops around this and it ran for 19 years until the plant shut down in 2004 (6,7,8).

The plant faced deadlines to implement statistical process control (SPC) to comply with ISO requirements, and I was tasked to implement this with automation. Meanwhile, I began learning about the new artificial intelligence technology of expert systems. At that time there was no suitable software to implement either SPC or expert systems as part of plant automation and control. The plant's existing mini-computer controls were limited in what could be added.

I began developing a tool on the more powerful Vax computers, drawing on strong software skills of others in the plant. Stumbles with my work on the catalyst state estimation control had taught me that only an operator-centric approach to automation would succeed. My goal was a control tool that would provide statistical process control; expert systems to decide how to respond to statistical shifts; in way that involves in and informs operators. After months of work, I commissioned live controls using PACE, the first implementation of expert systems into a general control tool with SPC, that could be configured to work against multiple process needs (9). This system included numerous inventions that would soon be patented.

1987: Tapping an opportunity to broaden my work outside of a single plant, I returned to 'HQ', where my expert system innovations soon earned a seat in DuPont's Artificial Intelligence Task Force, a small group of leading technology innovators with a charter to apply AI throughout the corporation. I continued my process control work, formalizing PACE into an application that could be widely distributed (9, 10). 

My expert system/process control innovations were recognized, and the company soon undertook to patent these with six US patent applications (11,12,13,14,15, 16). I began a career-long effort, as a proponent of the value, novelty, and strength of these innovations. I worked in depth with the attorneys prosecuting, drafting parts of the applications, and working closely during prosecution, writing responses to USPT office actions and visiting examiners to argue for allowance.

Working with a wider spectrum of DuPont manufacturing plants, I realized the potential to extend these expert system innovations to batch process. I made enhancements to PACE and we then filed a seventh patent application (17). At this point my patent expertise allowed me to draft a good part of the application and collaborate in writing claims, as well.

1990: Others on the AI Task Force were trying neural networks for paint color design, a data-intensive AI technology which contrasted with the knowledge-intensive nature of expert systems. This sparked a series of thoughts for me. I had already quite extensive experience with the process data collection systems used in DuPont plants, and I realized  that tapping these data stores for use in building neural networks could lead to significant process control improvement. I outlined the functions that would be needed to integrate neural networks with the data systems and the control systems, and I prototyped the develop interfaces to enable it. Already having allowances on most of the my expert system patent applications, DuPont management quickly accepted my proposal to patent these innovations.

This led to a significant effort over almost a year, working with attorneys to describe and support innovations that ended up spanning seven patent applications, which by the time of issue became ten patents (18-27).

1991: I was asked to join the Process Control Technology Center in DuPont Chemicals. This centralized resource made specialists available across DuPont's multi-billion diverse businesses in chemicals and polymer intermediates. The corporation formed a team to design a system to make neural networks easily used in manufacturing and control, and I was named to co-lead this team with another distinguished process control technologist. In tandem, I began development of production software for the data mining from DuPont's process database systems. I used this software in its early form to build and test a number of neural networks for process control or process monitoring. As our team finalized its design guidance, I developed the remaining control functions, completing ASPECT, the second major software system in my career (28). I continued a small amount of direct plant support (7, 9)

1993: In a turbulent corporate restructuring, I was moved to DuPont Information Systems, a reflection of my shared expertise in computers and software along with chemical processes and control. I was tasked to transition support of PACE to new contract support professionals, while continuing to evolve ASPECT and teach others how to apply this technology.

1995: In 1995 I was asked to join DuPont Engineering's Process Control Consultants, where I shifted focus to working on control improvements in DuPont plants. Polyester intermediates absorbed much of my time as they sought to survive in an increasingly competitive market. From 1995 through 2004, I designed and implemented 10 major control improvements in this nearly 1 billion lb/yr plant. These projects: 1995 MeOAc Stripper 1996 – TC Btms cooler optzn (29); 1998 Dehy control; 1999 Steam Header Optzn; 2000 – Feed Mix Metals Control (8); 2001 Vent Cond Load Control; 2001 – TC OH comp control; 2002 Me:Br Control (30); 2002 Oxidation Optimization (31). It was also around 1995 that I began training in figure skating.

Dynamic simulation was a substantial part of my work. Initiallyd I used Speedup to attack large non-linear problems, including the acetic acid/water distillation (dehy) and the simulation of the polymerizations in Terathane(tm) synthesis. But as Hysys became available, I adopted this tool with it's superior interface. The vendor, Hyprotech asked me to speak on the use of this tool for control studies (32). My polymerization work in grad school formed the basis for a successful simulation of Terathane synthesis.

Around 1995 DuPont released the patents on my inventions - now 17 in the US - to be licensed. Over the course of the next four years, I played a key role in presenting my inventions and the software that embodied them to potential licensing companies (33). In 1999 Pavilion Technologies in Austin licensed the portfolio with an option to buy an assignment. In 2000 they executed the option and paid the first installment.

2002:In 2002, I started a parallel endeavor and enrolled in a graduate certificate program in sport psychology as an evening study commitment. Around the same time, Pavilion signed me as a technology expert on their litigation team asserting infringement by Computer Associates of some of my patented inventions .

Along with a significant time commitment to the Pavilion litigation support, I continued work in polyester intermediates. After decades of operator control, we commissioned controls that optimized oxidation conditions, driving the reaction with the lowest cost combination of temperature, catalyst, and promoter.

During this time, DuPont was selling and shutting down large segments of the company based on continuous chemical processes. Demand dropped for the centralized process control services of which I was a part. My effort shifted into other businesses and other types of work. I developed simulations of a batch process expansion in Rosenburg, Netherlands, and spent three weeks in Honeywell's Amsterdam offices testing the controls against the simulation. I developed a global material and product flow database and diagram for management to better understand one of DuPont's globally distributed specialty polymer businesses. 

2004: In early 2004, DuPont announced yet another corporate downsizing. With the continuing decline in the company's need for skills like mine, I opted to take the separation package an leave the company. About the same time I finished my graduate certificate in sport psychology - work that amounts to roughly half of a master's program.

After 22 years of corporate life, I was ready for a change. Relocating would almost certainly force the end of my substantial custody responsibility for my two daughters, so moving was something I'd only do as a last resort. I started in parallel two efforts: to build a sport psychology coaching practice; and to publish and sell two books I had written over the past several years (34, 35). I made a small foray into sports performance (41,42), but it soon became clear that a sport psych practice was going to take years of hard work to build.

Meanwhile, my skills in web site design and my willingness to finally follow in my father's footsteps (a career salesman and manager of sales people) led to a surprising internet based bookselling business. As author, publisher, and bookseller, the margins were attractive. I put sport psych aside, and focused my efforts here. I wrote another book, then another, eventually totaling seven.

Off-the-shelf software for e-retailing was not serving very well, and I began developing a web site coded in ASP.NET/visual basic. In time I installed a server and began operating my own web server. An email list is a critical part of internet sales and marketing, and I began developing a system to handle a large list and comply with anti-spam guidelines. Moreover, having something valuable to send in email is essential, and over time I developed a set of about 90 mailings; then developed an automated system to send these each day, cycling through the set. Currently over 10,000 people receive a daily email.

2006: Pavilion's infringement suit against Emerson Process Automation was advancing into the Markman phase, arguing the definition of specific phrases used in the claims. Pavilion contracted with me to provide expert consulting to this effort. Later this progressed and I was asked to testify as an expert. I was deposed, then testified in a Markman hearing.

2007: As the case approached trial, I agreed to act the primary expert arguing validity. The legal team continued a search for an expert to argue infringement; after some time they came back to me and I agreed to also serve as the primary infringement expert. Midway through report preparation, the case "settled" as Pavilion was acquired by Rockwell Automation.  During this year, I also spent time preparing, then sat for the first of two exams in the process to become a registered professional engineer (PE). I passed the fundamentals exam and became a registered Engineer Intern.

2008: The once-settled case returned to life, and was called back. I was deposed at length as both the validity and infringement expert. As such, I was to be the primary witness in Pavilion's case. As the trial date approached, I received two days of coaching in jury testimony. The case settled the day before trial.

As a very small but stable business, I restructured the customer-facing operations into a Pennsylvania corporation. As an individual, I contract with the corporation to provide web site support and IT infrastructure and support.

2009: 2009 brought the publication of my sixth book. New data security standards for systems processing credit/debit cards made a server re-design necessary.

2010: New data security standards for systems processing credit/debit cards made a server re-design necessary. A new web server with a separate mail server went into service in 2010. Enhancements to the mailing list server improved delivery and reduced complaints.

2012: Looking forward to rejoining industry in chemical engineering, I decided that I wanted a recent project in hand. I decided to do an independent research study for publication. Although much of my industrial work was based on operating plant data in historians and made use of commercial simulation tools like Hysys, neither of these were available to me an an individual. I chose a problem I first encountered in DuPont Engineering: trying to compute from reactor kinetic modeling the molecular weight distribution resulting from the acid initiated polymerization of THF to make PTMEG. This is an industrially important problem, the basic was published in the literature, and DuPont no longer operates using this chemistry.

I began the work in the spring of 2012. It begins with polymer chemistry, as published by researchers from DuPont, from Japan, and from Europe. My standard was high. I wanted to clearly and completely understand the basis in chemistry for the model. This proved difficult, partly because I am not a polymer chemist, but also because this particular system is complicated. A debate went on in published articles, disputing the nature of the chemical mechanism. After too may trips to U of Delaware's library, I finally resolved that DuPont's Pruckmayr and Wu has indeed identified the chemical mechanism, and it is complicated.

Rather than a simple "Living" polymerization started by a proton, both proton and its counter ion are reactive. This leads to complex combinations of chain end groups, some of which are further reactive. To model it, nine chain species have to be included. These have 3 possible end groups of end the chain. Unlike a simple living system, some of these counter-ion reactions are chain terminating; and some transfer chain activity to initiator.

After many difficult trials, I finally correctly expressed all the kinetics in material balance equations, and solved the model as a system of ordinary differential equations. Using a lumped approach I found that the conversion vs. time profile was almost square, not curved. Studying carefully, I realized that the lumping inherently included the reverse activation reaction - which I specially assumed to be irreversible. Thus it seems a full distributed accounting would be needed to properly exclude that reverse reaction.

Again after a number of difficult trials, I got the balances right, and voila! The conversion profile matched the experiments. The distributed model reproduced all the time profiles, plus it gives a full molecular weight distribution. The question, though, of why the rates constants for this didn't match the published numbers required further lengthy literature study. Eventually the papers revealed that a lumped model was used to fit the published rate constants, so they represented a system with reverse initiation. A final question: why was the Flory distribution talked about in reference to this system. Yet another descent in the literature, this time going all the way back to Flory in 1936, revealed that no one had discovered an analytical distribution that represents the kinetics in this system.

2013: Finally by year end I had a manuscript in hand and asked for suggestions from several colleagues. There were no concerns about the technical work, but it seemed that my paper needed to reordered, and reformatted for a different journal. After editing and building figures and tables, I sent the paper to Chemical Engineering Science in May 2013. As of this writing, it is still under review, but I expect it will be accepted and should be online by later this year.

2015: I completed and published my seventh book. In the early fall I spent two months preparing for the PE exam in chemical engineering. Revisiting a number of design areas that I hadn't touched since my formal education was interesting and reassuring. With some refreshing, I felt confident and clear about the dimensions of chemical engineering covered in the exam. I sat for the exam at the end of October and passed.

2016: I January, I became a licensed Professional Engineer in Delaware.

 

 

 

It's a long i like the sky - Sky´ rick