COLI Meeting Summary- International Forum of Process Analytical Chemistry (IFPAC) Conference, January 23, 2002

INTRODUCTIONS:

End-Users:

Dave Marrow- Exxon Mobil: sees large deficiencies in current real-time prediction software for process analytics, has experience with a specific real-time prediction solution using Orbital/Analect hardware; is very interested to hear other end users’ opinions regarding what’s missing in current vendor offerings
Chuck Miller – DuPont: concerned about time/resources required to develop and support custom real-time solutions, has experience with specific real-time prediction solution using Orbital/Analect and Foss/NIR hardware
Aerenton Bueno, Petrobas: works with NIR/gasoline applocations, is familiar with Pirouette, others in the company have Unscrambler
Steve Doherty, Pharmacia: concerned about compliance of solutions in highly-regulatory environments
Frank Schweighardt, Air Products: interested in on-line chemometrics applied to medical products
Troy Francisco, Honeywell: has dealt with many different types of analyzers and software, is frustrated with difficulties in implementing chemometrics on-line
Don Young, Chevron

Analyzer Vendors:

Bruce McIntosh. Hamilton Sundstrand: system design and applications for Orbital/Analect process spectrometers, on-line “PC80” operating software
Mike Doyle, Axiom Analytical: in the process of developing an instrument-independent software package for on-line analysis
Blayne Hirsche, Axiom Analytical
Mary Thomson, Remspec: primarily working with fiber-optic probe applications, experiencing stronger demand for real-time chemometrics applications
Mike Power, Convergent Solutions
Sharon Deram, SBI Analytical
Joe Slater, Kaiser Optical Systems

Chemometrics Vendors:

Neal Gallagher, Eigenvector: Chemical Engineering background, teaches short courses, software training for PLS tool box for MATLAB, has a strong interest in facilitating real-time applications of chemometric models
Brian Rohrback, Infometrix: developers of Pirouette, includes software development, training, and custom software systems

Academia/Government:

James Whetstone, NIST, Chemical Science and Engineering Lab: interested in standards for process analytical industries
Steve Brown, University of Delaware develops new methods for calibration, classification, and method transfer, wants to see his new tools used, but not abused
Paul Gemperline, East Carolina University: develops novel algorithms and software, wants his tools to be used in the real world

DISCUSSION:

Where we are, and where we’re going:

Dave Marrow began the discussion by noting that this initiative, which is supported by CPAC, is still rather new<. The purpose of this meeting, and other such meetings throughout the year, is to generate interest and solicit feedback on the on the concept of facilitating “open architecture” on-line chemometrics software solutions. It is hoped that we will be soon be able to generate a “critical mass” of interest and participation, which will enable this initiative to become self-sustaining.

The organization of this initiative involves several different entities:

The Sponsor: CPAC (Center for Process Analytical Chemistry), who provide sponsorship and support services (web site, etc.) for the initiative
The Stewards: Dave Marrow (ExxonMobil) and Chuck Miller (DuPont), who facilitate the development of the initiative
The Steering Team: consists of about 15 persons representing the four interest groups in this initiative (end-users, analyzer vendors, chemometrics vendors, and academia); has the “final word” on all decisions made for the initiative.
Participants: anyone who is interested in providing input and direction to the initiative.

At this time, as the initiative seems to be gaining momentum, it is important that we define its identity and scope. This needs to be defined in terms of several factors:

WHY this initiative?
WHAT….will be the specific deliverables of the initiative. What should this initiative NOT do?
HOW: What specific IT tools could be used to make these deliverables? How do we protect intellectual property rights of the vendor companies?
WHO….could be useful resources for the initiative?
WHERE…could solutions developed by this initiative be applied?

With these questions in mind, the discussion was opened.

Open Discussion:

Neal G.: his IFPAC talk summarizes the needs for on-line chemometrics; must have some kind of standard – a file standard or a communication standard (like Active-X); such standards currently exist in the process control/DCS world- could we leverage their knowledge?

Bruce M.: as an instrument vendor, is rather nervous about the direction of the initiative; are we to offer vendors a set of “rules” on how to interface with another prediction module? Would rather have a standard “library” of Real-time algorithms; also sees intellectual property issues with “homemade” algorithms

Dave M: ExxonMobil has “homemade” methods, but needs a common framework on which to implement them

Mike P.: can develop a “hierarchy” of data processing routines

Bruce M: another layer of prediction methods triggered by control events?

Neal G: lots of different data pre-processing methods to consider as well; end users are concerned that software solutions will not be “open source”

Brian R.: it will be a challenge of this initiative to preserve the intellectual property/competitive position of vendor companies

Paul G.: suggests adopting an object-oriented approach: “object” contains both model and preprocessing parameters, along with code that expresses the model; end user can assemble object in any order; however, will need an “engine” that links such objects together, as well as an individual to build the objects; also, would need specifications for the objects.

Bruce M.: should we build CFR-11 (regulatory data flow restrictions) into any of our standards?

Dave M.: regulatory issues should not have an impact on our final objective

Neal G: we should keep in touch with other CPAC initiatives: connectivity (CONNi)- developments in OPC, and modular sampling systems (NeSSI) – sensor fusion applications

Mary T.: an easier on-line chemometrics package would enable easier model “robustness testing”

Bruce M.: a standard for data format and algorithms might be of value, but need to consider audit trail issues for applications in regulated industries

Steve B: standards are easy to set up, once their objectives are defined

Paul G: model parameters can be exported with embedded code as a single object; object can be executed in Matlab, C++, etc…; software validation can be up to the individual vendors.

Brian R.: some folks sell models to others, and don’t want to disclose calibration data- this is not an issue unless the calibration data is needed to execute the model (for example, in LWR)

Mike P.: need a well-defined model diagnostic metric; there are several out there already- for example, some use a “Q” value, while others use prediction “residual” or “Mahalanobis Distance”

Paul G: I/O standard; will need to allow modeler to specify “objects” to be included in model output; like model diagnostics, there are a lot of different statistical figures of merit! Could specify a “core” of recommended model outputs

Neal: NIST had done some testing of statistical packages- perhaps we could use data, standards from their testing to test/compare different on-line chemometrics approaches?

Chuck M.: any limitations of object-oriented solutions? (for example, operating systems?)

Paul G.: if active-X is used to pass objects, then operating system would not be an issue

Brian R.: has economic concerns regarding this initiative- it could make existing vendor’s products less-valuable

Paul G.: does not agree- sees lot of value in a good user-interface for both development and implementation

Brian R: but some vendors already have products for both (for example, Informetrix’s Pirouette and In-Step)

Brian R.: we will need to set up a way to get analyzer data into chemometrics packages- is this really an issue anymore?

Dave M: should we develop another JCAMP? (big sigh from the audience!)

Brian R: “Net CDF” (??) is the most global solution, but is more suited for software engineers, and more of a “black box”- not really “open source”

Dave M: 4 questions regarding standard data formats:

Is one required in order for COLI to meet its objectives?
Can we reasonably expect to be able to deliver a specification for a standard data format?
Is the Mathworks .MAT an acceptable format?
What other possible formats are available (we would prefer to use an existing format rather than to develop a new format!)

Post-Meeting Discussion: Name Change?

After the meeting, Dave M., Neal G., and Chuck M. discussed the possibility of changing the name of the initiative. It was thought that the “On-Line” part of the name “Chemometrics On-Line Initiative (COLI)” can give the wrong impression of an internet-based initiative. In addition, there’s a concern about the negative connotations of the “COLI” acronym. We agreed that some mention of the words “Process Analytics” in the title would provide a more accurate description of the initiative. Based on this, we came up with the title “Chemometrics for On-Line Process Analytics (COPA)”. Unless we hear any strong objections from you, we will use this new name and acronym starting after the Pittcon meeting.

Attendees at COLI Discussion- IFPAC 2002 (at least, those who signed the register..):

Neal B. Gallagher; PO Box 561; Manson, WA 98831; Eigenvector Research, Inc.; 509-687-1039
: Brian Rohrback; Infometrix, Inc.; PO Box 1528; Woodinville, WA 98072; 425-402-1450; Bruce McIntosh; Hamilton Sunstrand Sensor Systems; 2711 N. Garey Ave.; Pomona, CA 91767; 909-593-3581 ex 6106; bruce.mcintosh@hs.utc.com; Mary Thomson; Remspec Corporation; 512 Leadmine Road; Sturbridge, MA 01566; 508-248-1462 ex 203; mat@remspec.com; Mike Doyle; Axiom Analytical, Inc.; 177510B Sky Park Circle; Irvine, CA 92614; 949-757-9300; mdoyle@goaxiom.com; Mike Power; Convergent Solutions; 235 E. 20^th Street; Idaho Falls, ID 83404; Blayne Hirsche; Axiom Analytical; 17751 Sky Park Circle #1B; Irvine, CA 92614; Bhrische@goaxiom.com; Aerenton F. Bueno; Petrobras; S Jose Dos Campos; Brazil; Aerenton@petrobras.com.br; Troy W. Francisco; Senior Staff Scientist; Process Analyhtical Chemistry; Fluorine Products; Specialty Materials; Honeywell; Buffalo Research Laboratory; 20 Peabody Street; Buffalo, NY 14210; 716-827-6368; 716-827-6236 fax; 888-538-2444 jpager; troy.francisco@honeywell.com; Frank K. Schweighardt; Manager; Process Measurement Technology; Air Products and Chemicals, Inc.; 7201 Hamilton Blvd.; Allentown, PA 18195-1501; 610-481-6683; 610-481-6517 fax; 800-759-8888 pin #1739487 pager; schweifk@apci.com; Donald C. Young; Staff Scientist; Integrated Laboratory Technologies-Petroleum Chemistry; Chevron Research and Technology Company; 100 Chevron Way; PO Box 1627; Richmond, CA 94802-0627; 510-242-3298; 510-242-1792 fax; dcyo@chevron.com; Steve Doherty (for Bob Roginski); Pharmacia Corp.; 4901 Searle Parkway; Skokie, IL 60077; Rtrogie@pharmacia.com; Sharon Deram; SBI Analytical; 70 Hickory Lane; Lincolnshire, IL 60069; Spderam@ix.netcom.com; Joe Slater; Kaiser Optical Systems; 371 Parklamp Plaza; Ann Arbor, MI 48103; Slater@kosi.com; Steven D. Brown; Chairman and Professor; University of Delaware; Chemistry & Biochemistry; Newark, Delaware 19716; 302-831-1247; 302-831-6335 fax; sdb@udel.edu; Paul J. Gemperline; Professor; East Carolina University; Department of Chemistry; 327 Falanagan Building; Greenville, NC 27858-4353; 252-328-6767; 252-328-6210; gemperlinep@mail.ecu.edu; Dave Marrow; ExxonMobil Chemical Company; Spec Core Process Control/Analytics; Baytown, TX; david.marrow@exxonmobil.com; 281-834-1789 office; 281-834-1509 fax; 281-380-8433; Chuck Miller; DuPont Engineering Technologies; 140 Cypress Station Drive; Houston TX 77090; 281-586-2545; 281-586-2504 (fax); charles.e.miller@usa.dupont.com