How Business Works: #236 in the Series

In 1976, we had just completed what turned out to be the first urban airshed simulation using a grid model with realistic photochemistry. By "realistic" I mean that it had a close to accurate estimation of the primary source of photo-oxidizing radicals (from carbonyl compounds rather than what was previously thought to be the primary source: oxygen atoms reacting with hydrocarbons), thermal decomposition of PAN, and a strong nitrogen oxide sink related to aromatic hydrocarbons.

One interesting thing about this particular simulation was that it did not involve Los Angeles. It just so happened that a project involving Denver had coincided with several model upgrades, including the new chemistry, so Denver got the goodies before LA did.

The project team consisted of a goodly fraction of the employees of the (rather small) research consulting firm that had originally won the main EPA follow-on work for developing a photochemical grid model: Systems Applications Inc., not to be confused with Science Applications Inc., or several other firms that went by the initials SAI. Systems Applications no longer exists as such, having been part of the merger and acquisitions whirligig in the 1980s, followed by a breakaway group going to start up a unit of Environ, though not many of those at SAI in 1976 are with either the ghost of SAI or Environ now. That's the biz, you know?

Anyway, part of the SAI business model was to do these government research and consulting gigs, which did not have much profit margin, followed by environmental impact work for other groups, usually corporate, which did have decent profit margins—sometimes. And thereby hangs this tale.

After the work for the Denver Regional Council of Governments (pronounced "Dr. Cog"), we got a request for an impact statement for a facility that had a natural gas turbine power source. Natural gas burns without much in the way of hydrocarbon emission, but the combustion temperature creates some nitrogen oxides, NOx in the lingo, and we were charged with determining the air quality impact. The thing only emitted a few kilograms of NOx per day or thereabouts, barely enough to register on the meter, as it were, but part of the song and dance of environmental impact statements is to do your "due diligence" and if you can get the cutting edge of science on your side, well, good on you and here's your permit.

I'd been the primary modeler on the DRCOG project, for a lot of reasons that I'll describe some other time, and there was a computer programmer/operator who worked with me, and a project manager above me. This was back in the days of punch cards and CDC 7600s, and pardon while I get all misty eyed, okay, that was plenty, because, really, feeding cards into card readers to run programs sucks.

I asked the programmer how much time he expected the job to take. The only thing that needed be done was to add one single point source to the point source input deck, then a bit of analysis, AKA subtracting several numbers from each other and maybe drawing a picture or two. He estimated the time at something like three days, but said, "Call it a week."

I knew how much a week costed out at, so I got the dollar figure, then doubled it, and reported that as my estimated cost of the project to the Denver Project lead. He doubled my estimate and gave it to the Comptroller.

The Comptroller doubled that number and gave it to the company President, who then doubled it and made that offer to the company that wanted to hire us. They signed without blinking.

Okay, so that's between 16 and 32 times what the programmer had expected the thing to cost, a nice profit margin, and good work if you can get it.

Then the programmer added the emissions to the program, ran it, and compared it to the original "base case" or "validation" simulation. They were the same.

Okay, really small emissions source. It's not surprising that the effect was minor, miniscule even. But he expected something. I think he was looking at like five or six digit accuracy in the printouts. There should have been some differences in the numerical noise at least. So he multiplied the source strength by ten, then by a hundred.

Still no difference.

Well, a programmer knows a bug when it bites him on the ass. He went into the code and found an array size limit that basically meant that any point source greater than #20 didn't get into the simulation. The impact source we were looking for had been added to the end of the list, so it didn't show up.

But.

The Denver region at that time had one major power plant that was responsible for something like 30%-40% of all the nitrogen oxides emitted into the Denver airshed. And, wouldn't you know it, that power plant was like, #45 on the list, or whatever. Higher than #20, that's for sure.

Oops.

So now we had to go back and redo our base case. We also had to redo every single simulation in our original study, and rewrite every report, and all the papers that were in progress, and notify the nice folks at DRCOG, who, it should be noted, had already paid us for all of the above when we did the original study, so they weren't about to pay us to do it again. We were lucky in one way: large, elevated point sources (like power plants) don't have nearly the impact of ground-based sources like automobiles, so the omission hadn't had that much effect on our original simulations, at least not near the air quality monitoring stations that we'd used to test the veracity of the model. There were some differences, of course, and tables changed, future impact projections were modified, etc. etc. Oh, and we got to use the original base case as a "what if" scenario, as in "What if Denver's largest point source of NOx emissions were switched off?"

Fortunately, we had some money to do all these things with: the environmental impact contract. I was told that we did actually wind up making a profit on it. I think it was in the low triple digits.