Modeling at PPG Industries - An Interview with Michael Makowski, Scientific Computing Group

Introduction

PPG Industries is a global supplier of coatings, glass, fiber glass, and chemicals. The company has about 50 production facilities in the United States and about 120 worldwide, including subsidiaries, joint ventures, and equity affiliates, employs more than 34,000 people, and had 2002 sales of $8.1 billion.

PPG is the world's largest producer of transportation coatings and is a leading maker of industrial and packaging coatings, aircraft transparencies, flat and fabricated glass, continuous-strand fiber glass, chlor-alkali and specialty chemicals, and architectural coatings.

The interview

What follows is the transcript of an interview conducted by Accelrys with Michael Makowski, Scientific Computing Group, PPG Industries.

1. What modeling, simulation, and/or informatics software does your company use?

For computational chemistry, we primarily use commercial products from Accelrys, MDL and Gaussian, although we also license some codes developed at National Labs (such as NWCHEM).

2. What do you use it for? How does this work fit in with your company's long-term goals?

We apply computational chemistry to a broad range of problems. Specifically for our coatings business we are interested in reaction thermochemistry (cure), ultimate film properties (toughness, hardness, modulus), miscibility and diffusion, interfacial phenomena, stabilization and degradation mechanisms (UV absorbers, HALS), new color development (specialty pigments), flow and leveling (appearance), surface effects (wetting, adhesion), and dispersant stability (pigment binding).

In our chemicals business we use modeling to study the coloration kinetics of photochromic dyes in polymeric matrices, OLED materials development, and reaction mechanisms for fine chemicals and pharmaceutical intermediates. In our glass business, a recent focus has been on studying photoactive TiO2 for our self-cleaning glass product sold under the SunClean brand.

Our approach is to use modeling to reduce discovery and development cycle times, gain fundamental understanding of our core technologies and competencies, gain a competitive advantage, and ultimately reduce costs. We view modeling from a continuum perspective that covers all length and time scales, and attempt to use any and all methods in conjunction with experimental data to solve a problem or add value.

3. Have you published work in the scientific literature and/or general press that uses computational software? If so, when and where?

Where possible, we have published or presented work in the open literature, but our primary vehicle for the disclosure of new developments has been with patent applications.

4. What did the software enable you to do that experimentation didn't?

In many instances, we've been able to study a problem where experimentation simply couldn't provide an answer, especially in the areas of mesoscale phenomenon and in developing structure-property models that involve electronic structure effects.

In many instances we've been able to develop structure-property relationships much faster than would be possible experimentally.

5. What would you say are the main scientific advantages of using computation over experimentation? Likewise financial advantages? Did its use save resources - i.e. time, money...?

Scientific advantages include the development of a fundamental understanding for a particular phenomenon rather than simply fitting an empirical model. We have even used modeling results for supporting patent applications based on first principles that have allowed us a broader coverage of claims.

Financial advantages have been clear in terms of time and materials savings in addition to steering efforts away from dead-end avenues.

6. How long would you say that it took for your company/organization to re-coup the initial investment in the software (including initial, installation and running costs) with any cost savings mentioned in the previous question?

Modeling has matured to a level here where we simply don't think in these terms. We view the overhead associated with modeling as an operating cost necessary to drive innovation in a modern R&D environment.

Within any one of our groups, it has been very easy to identify at least one project that has paid for the costs of modeling several times over.

7. Did the use of the computational chemistry techniques result directly in refinements to existing processes? And, if so, how much has it saved your company? And in the future?

Most of our successes in compchem have not been in the area of process improvement, but rather in product improvements, optimisations or new discovery.

8. Did the use of the computational chemistry techniques allow you to gain a competitive advantage?

We have several commercial products that computational chemistry has played some role along the development cycle. Competitive advantage has come in being first to market and/or developing broad patent strategies to protect our intellectual property.

9. What do you and your organization plan to use the software for in the future?

We have seen an increase in mesoscale applications and have discussed the possibility of incorporating informatics strategies as well.

10. Would you recommend the use of modeling/simulation to your peers?

A major component of our mission statement is the inclusion of a 'Modeling Advocacy and Education Program', so the answer here is certainly "yes".

11. Are you authorized to speak to journalists on behalf of your organization?

Yes.

12. Likewise, would you be willing to be interviewed by a reporter about your use of our products?

Certainly.