1. Thermal newsletter- illustrating the goals for this function and ways to help product development
Just keep it fun, and be wary of going too technical. Most engineers don't understand entropy, thus don't go confusing others with it. Lots on pictures, few words...
If there is already a site newsletter, I found this worked better because a) the person putting together the paper always battle to find good articles, and generally jump at the chance to get one or 2 technical articles as well and b) the circulation is a lot wider than a technical paper and more importantly c) you don't need to do the marketing of the paper - and thus can focus on more important thermal issues.
2. Interview middle and upper management on the goals they would like to see met from a thermal standpoint.
Mostly, you will find that management do not know what they would like to achieve from a thermal standpoint. What I found worked better was to go with a proposal on what you think shoould be achieved, and ask for their comments/ support - have the robot chart I talked about in your back pocket. Just make sure there are lots of dollar signs on the proposals - even if it is rough assumed figures. Even a statement like "we believe we could have a fuel saving of around $10000 per year" is a lot better than saying "we can increase the efficiency by 0.05%".
3. I like your training idea. Maybe I could set up a short intranet class on thermal analysis and its benefits
Keep at it - there is a lot of good material out there. Just make sure you turn it into something plant specific. Also remember my earlier comment of not having any mathematics in there!
4. Write papers (internal and external) on useful problems solved.
Focus on external papers. I'll probably get stoned for this by most engineers, but the sole purpose of these papers are to make your management look good. After all your intention is to gain support over the long term. Even if management stuffed up 49 out of the 50 thermal areas, focus on the one which was managed well, and write the article about that.
One last trick, which most engineers often overlook:
Use the right unit of measurement. If you e.g. want to say the dissolved oxygen is good, use 2 parts per million. If you want to say its bad, use 2000 parts per billion.
Enjoy!