top of page
  • bjones148

Tips from a modeler on how to work with a modeler

Once upon a time I had somebody approach me about doing some work for them. Being a good consultant I immediately said sure, “I’d be glad to help.” However, as I tried to put a proposal together, I couldn’t get clear information as to what exactly they wanted—I got vague generalities like “the plant will be located in such-and-such a town,” or “I need you to model air emissions associated with this type of a facility.”

That experience got me reflecting on how a client can best work with a modeler. So I decided to take a bit of my snowy afternoon before the Super Bowl to offer up a few thoughts.

Thought No. 1: What is the purpose of the modeling?

If I understand the purpose of the modeling that goes a long way toward helping me understand what needs to be done. There’s a very big difference in my approach if the modeling is being done to get a permit or if it is being done to answer a theoretical question like “can we increase our throughput without triggering certain regulatory actions.” If the conversation starts with “I am building a new process line and the state has told me we need to model” then I know that we need an analysis that demonstrates compliance with applicable standards in as clear-cut a fashion as possible, because the more “artsy” we get with the analysis the more opportunity there will be for scrutiny from the regulator which will slow down the review time.

So don’t think of your modeler as a mere black box—bring them in early in the process and let them be part of your business decision. Tell them what your goal is and let them use that understanding to steer the modeling project in a direction that is most helpful to you.

Thought No. 2: Provide accurate, specific information

There’s no question that modeling can be very much an art, but at its core modeling is a science—and good science needs good data.

There are three basic kinds of data needed to do modeling: source information, receptor information, and meteorological data. Most of the time the meteorological data are going to be completely in the realm of the modeler, so you likely don’t have to worry about that. On the receptor side you’ll need to provide an accurate description of the ambient air boundary—if you give unclear information the modeler might end up putting receptors in a location that is not truly ambient air, and before you know it the model is showing exceedances there and then you’re going down the rabbit hole of taking operational limits to fix a “problem”—but in reality that problem doesn’t exist because you have a receptor in the wrong spot.

A bigger issue is on the source information side. I’ve had clients hand me model inputs from their last analysis along with very clear instructions like “double the emissions from source such-and-such.” Clear guidance like that is terrific, and there’s not much chance of me misinterpreting that model setup.

At the other end of the spectrum, I’ve had clients hand me a pile of paper and say “what you need is in here.” That usually leads to me spending a lot of time (and their money) wandering through it, and invariably I end up coming back to them and saying “OK I found this information but I still don’t know that information.” But beyond just being an inefficient way to work through this part of setting up a modeling analysis, if all I have are nebulous instructions from the client there’s a chance I’m missing something completely.

So to not only make sure that your modeling analysis is done efficiently but also properly, be as clear and specific as you can be with your modeler when providing source information. Give them a map that indicates where all emission sources are along with the ambient air boundary—just don’t give them a street address that they can look up on Google Earth and then try to guess where the sources are and the fenceline is. Give them the information they need to calculate emissions, from the sizes of equipment to fuel throughputs to operational schedules. And finally give them source parameters like heights, diameters, and flows—vendor documentation is a great place to get that kind of information.

Thought No. 3: Ask good questions

Once the modeling results start coming in be sure to ask your modeler about them to make sure you understand them. Where are the hot spots and how big are they? It could be that a modeling result is limiting your operation, but if you check with your modeler you may learn that it’s a single, isolated spot that is on your property but just not fenced in—so for a small amount of additional fencing you might be able to make that controlling concentration go away which would enable you to increase production.

Another good question to ask is which source is (or which sources are) contributing the most to the controlling concentration? It could be that the source driving your controlling impacts has its emissions calculated in a very conservative manner, and all you have to do is back off of that conservatism to lower that controlling concentration which could open up some operational flexibility.

To sum up my thoughts:

  • Use the modeler as a business partner, not just a black box

  • Give the modeler the best information you can up front

  • Work with the modeler to make sure you understand the results in the context of your operation

If you can do this, you’re going to get the best modeling results you can, and you may even become a hero to the operations folk in your company by being able to tell them “I worked it out with our modeler such that we can make more product than we thought.”

31 views0 comments


bottom of page