As promised in earlier posts, I am going to elaborate a bit more on best practices for succesful implementation of computer simulations in the engineering industry. Obviously, a really successul implementation (e.g.  idea for a new engine =>computer simulation=>  new engine as a company product) is a bit more involved than what is described by the steps below. At MVP Modeling Solutions, we give training seminars that can be several days long and sometime stretch out into months of consulting in order to implement the best practices for computer simulation in the most effective way possible.

But, if you keep in mind even some of the steps outlined below when getting ready to choose your next modeling software tool, they will help guide you in selecting the best software tools possible for your specific problem. The focus here is on reactive flows, but general versions of these questions can be applied to CFD and FE modeling as well.

Evaluation Criteria for Reactive Flow Modeling Software

1) Identify the Problem:

The very first step is to decide what problem you are trying to solve. If the overall goal is to design a new car engine, for example, the first step is to break down the problem into manageable pieces that can each be more or less accurately represented by a computer model. Below is a list of questions that you can ask in order to identify what those manageable pieces might be.

1. What is the overall goal? (e.g. in the case of a new engine design idea - to help design a new automotive gasoline engine that would have greater fuel efficiency and lower NOx production)
2. What part do you personaly play in achieving this goal? (e.g. I am a design engineer who is designing the combustor part of the engine)
3. What variables can you control in this process and what is outside of your control but that you need to take into consideration? (e.g. I can control the flow rate of the fuel, number and location of fuel injectors, and fuel to oxidizer ratio. I have to work with a specific mixture for gasoline, thus have to work with certain chemistry or the fuel. I have some geometric constraints since the size of the combustor has to be within certain specifications.)
4. What are the important physical and chemical processes that you need to consider? (e.g. In case of the combustor, both efficiency and NOx (both part of my overall goal) are controlled by physical processes – mainly mixing, turbulence and diffusion and chemical – reactions of fuel and oxidizer. )
5. What assumptions can you safely make in order to achieve your goal? (e.g. I can assume that the gas-phase reactions in the combustor section are much more important for the NOx generation and efficiency than any surface reactions that might happen between the walls and gaseous species. I assume that after the fuel has been sprayed into combustor it vaporizes so quickly that the phase-change does not affect NOx production.)
6. Based on your overall goal in step 1 and assumptions you made in step 5, is it possible to separate some of the physical and chemical processes in step 4 and study them individually? (I could study chemistry to see how NOx production and temperature inside the combustor are affected by chemical reactions only. I can also separately study mixing and turbulence effects to see how fast the fuel molecules and oxidizer molecules are brought into contact with each other. For example, if this process is almost instantaneous throughout the combustor, I can assume chemistry dominates and not worry too much about modeling turbulence effects. But if I there are sections in my combustor that perhaps are more fuel-rich than others, temperature, chemistry and therefore NOx production and efficiency of the engine will be effected by turbulence and mixing and I need to understand these effects as well.)
7. Pick the best software available for each separate process that you identified in step 6. Typically, the more physical and chemical processes that the computer software claims to model, the less accurate the resultswill be and/or the longer the simulation will have to run.  So how do you choose the best tool?

2) Identify the Software Tool(s):

1. Identify your personal software needs:
   1. What is your programming experience? Do you have time, energy and expertise to get into the nitty-gritty of understanding, re-writing and correcting unsupported code? If yes, software developed through academia or your company developed in-house software might be an option. Typically, it will not require any money upfront, although the time you spent on customizing and de-bugging the code can easily exceed thousands of dollars in company’s man-hours.
   2. Are you busy conducting research and need commercially supported software that would provide you with condensed, focused training, tutorials and manuals? The downside of commercial software is that you need money, in most cases a lot of it, to be able to purchase the software packages you need. For some CFD packages, the investement can be upwards of thousands or hundreds of thousands of dollars.

1. What exactly do you want the computer simulation to do for you? It is best to be as specific as possible. Refer to your answers in Identifying the Problem section to answer the following:
   1. What specific physical processes do you need the software to model? Can you determine how those processes affect each other or is that one of the variables that the software will need to calculate? In case of the automotive engine, I could use software like CHEMKIN to determine everything I can possibly need to know about the chemical-kinetic interactions of fuel, oxidizer and radical species and use a CDF code, such as Fluent, for example, to study turbulent and mixing effects of the injection sprays.
   2. What are your input variables and which variables you need the software to calculate for you? For example, I might be able to control flow rate, inlet temperature of the fuel, and configuration of the inlet nozzles. I need the software to tell me the concentration of NO2 and NO species and conversion ration of fuel to combustion products.
   3. What assumptions can you make about your system? Refer to step 5 in Identifying the Problem section. Identifying as many assumptions as possible will go a long way in helping you find the right modeling tools.

1. Finally, take a look at the software needs of your group and/or company.
   1. Does your research group tend to use the same modeling platform?
   2. Is there particular software program that your group or company seems to favor?
   3. Will there be future research projects that might use the same software?
   4. Talk to other groups in your company that are looking for computer modeling. What are their needs? You might find that your interests are very similar and you can combine group budgets to purchase the best software for your needs.

Summary:

Define your problem the best you can. Write down all of the known and unknown variables, assumptions and specific quantities that your modeling experiment will need to determine.

Put together a systematic process for evaluating various software options available. Make sure that you are comparing the same processes, on the same operating systems, and same computers when you run various codes.

Consider things like, desired output, assumptions made, documentation/training available, support available, and future needs of your group/company when selecting modeling software.

As a final note - few companies realize how much money they can actually save, by adding more carefully selected computer simulation experiments into their engineering and R&D processes. From first-hand experience, I can tell you that these savings can range from tens of thousands to millions of dollars.

Interested in saving money for your department or company? Give us a call and I will be happy to set up an absolutely FREE phone consultation with you to discuss how your company can conduct more succesful computer simulations, while saving $$$ at the same time.

All the best,

Masha

P.S. Just a reminder, MVP Modeling Solutions is not affilited with any one software company in particular. Our goal is to promote computer simulation as means to develop better engineering products, cheaper and faster.

P.P.S. As always, blog readers, myself included, would love to hear your thoughts on the topic! Post comment below.

This entry was posted on Wednesday, June 10th, 2009 at 4:41 pm and is filed under Computer Modeling Best Practices. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.