INFORMS 2008, Part II
Here is more from the INFORMS annual meeting:
- Misha Adamantiades talked about the EPAs IPM electric power model. This is a massive LP model of 300,000+ constraints, around 4.5 million variables, taking 4-8 hrs per run. One thing it is used for is to study the impact of various policy enactments, such as greenhouse gas cap and trade proposals. So given a particular bill that may be pending, the model can be run to determine what generation mix will result, among many other outputs. The speaker referred to the fact that most emissions in the US come from the electric power industry, while at the same time clean technologies and programs in that sector typically result in the greatest emissions reductions per dollar spent. For more on IPM, see the site.
- The “Doing Good with Good OR Plenary” on energy and the environment was given by Philip Sharp of Resources for the Future, an energy think tank. Among the work they have done is consulting on the RGGI auctions. He talked about the challenge to analysts caused by the difficulty of getting a correct fix on resources, markets, etc. For instance, future estimates of supplies of gas and oil have fluctuated significantly over the years. He was optimistic about the role of OR analysts in tackling these challenges. His talk was followed by a panel discussion. The first speaker, Brian McLean of the EPA’s Office of Atmospheric Programs, discussed the successful acid rain reduction program. It made use of LP models aimed at minimizing cost. The cap and trade program for SO2 has had a significant reduction impact and ended up costing less than expected. He showed some striking maps of the improvements. The potential health benefits were not fully studied during the planning, but there was time to do so later on, and they were significant as well. The program has been an iterative process going through planning, implementation, assessment, further planning and so on.
During the Q&A after the panelists had presented, it seemed there was some kind of misunderstanding or difference of opinion between the panelists and at least some of the audience that was not reconciled by the conclusion. It stemmed from statements by more than one panelist about the importance of OR analysts not advocating. I do not recall all that was said, but (from some of the panel) it almost sounded like if one could do good with good OR on the energy/environment front, that was only because doing good coincided with minimizing cost. The questions from the audience mentioned young, future OR analysts “wanting to make a difference” or the OR analyst working to improve the environment because that was part of that person’s “worldview”. The panelists were mostly government-based, or consultants to the government, so maybe that was what was guiding their view. That is, the decision-maker should be the one deciding on what is to be optimized, so leave that to them, and leave the math to the analyst. There is something to be said for that. Maybe there is a danger of losing objectivity if your model results do not agree with your environmental perspective. Of course, you can never truly divorce yourself from your views, even if you think you can. But should you have to? A continuation of this discussion would have been very interesting. Were you there? What do you think?
- In another session, Yihsu Chen of UC Merced gave a talk on CO2 leakage and emissions spillover associated with the Regional Greenhouse Gas Initiative (RGGI). The idea as I understood it is that a regional emissions tax will raise the cost of electricity produced in the RGGI states. As a result, utilities will start to purchase cheaper power from neighboring states outside of the region. The example he showed was power coming from PA, which is not a RGGI state, to MD, which is. So the increased demand for power from PA, much of which happens to be from high-CO2-emitting coal, leads to greater emissions than were expected. Another negative side-effect is that coal plants also produce SO2 and NOx, so pollution of those will increase as well. One would think the RGGI group would have thought of this (see this document), and Chen did cite a 2007 reference from the group suggesting much lower leakage. But he appears to be claiming his analysis is more accurate. This could all become moot quite soon if a nationwide system is implemented, as many think it will be.
- Lastly, Max Henrion, CEO of Lumina Decision Systems spoke about data center efficiency. He used Lumina’s tool Analytica, a visual tool for decision models. A page on Lumina’s web site touts Analytica’s role in a number of green applications from green building design to climate change modeling. Update: A review of Analytica 4.1 in the June 2008 issue of ORMS Today gives a good overview of the program.