|
|
|
|
|
01/06/2011
 Email to a friend
 Comment on this article
|
Ian Adcock talks with Reaction Design CEO Bernie Rosenthal about the company's groundbreaking computational fluid dynamics package for combustion analysis
 Even after more than 100 years of development, and the massive steps forward in sophisticated software and increasingly powerful computers, the ability to predict accurately what happens when fuel is ignited in a combustion chamber is still something of a mystery to engine designers. "We've seen a requirement storm brewing for a few years now," says Reaction Design's CEO Bernie Rosenthal. "Not just climate change, but also financial considerations, materials security and health concerns. That's driven requirements down to engine designers to comply with tighter emissions regulations, including particle size, changing fuel combinations and global differences in the make-up of certain fuels, as well as future fuels such as second and third generation bio-fuels."
Rosenthal and his team "see the engine as a chemical power plant, where you basically mix fuel and air, and get the chemical combinations of CO2 or NOx, particulate matter, unburnt hydrocarbons, etc, as a result." Before they started developing the new FORTÉ Computational Fluid Dynamics (CFD) package, they went to the engineering community to discover what challenges designers faced with available software packages. "They asked us for three things: accuracy, as they couldn't reliably predict ignition or the effect of different fuels, never mind emissions. Some of the models being used require calibration, which basically means you almost need to have built an engine before you can figure out if you've modelled it correctly, which kind of defeats the purpose of modelling.
Secondly, they asked for faster time to solution – the overall time spent by the engineer from set-up of the simulation through to visualising the results. These guys are under enormous pressure to get new engines to market. The subtext was that they were generating input for simulation, which was taking too long. And, lastly, they were looking for a more intuitive design flow where they didn't need multiple tools to do their job and they were able to reliably walk away from the simulation overnight, allow it to work and return in the morning to find the result." What engineers wanted was predictive solutions, rather than an explanation of what had been built.
Commercial approaches, such as those employing red cued and simplified models, or populating tables that use software to deliver the answers, weren't sufficient. "We saw people dealing with this in nonreactive flow simulations that didn't take into account any of the fuelmixing effects or the combinations of the chemistry in the engine. Basically, looking at the turbulent flow inside the cylinder and then using experimental testing to try to understand how different fuels might affect that behaviour and, lastly, we saw people saying they will deal with this in the after-treatment system," says Rosenthal.
Effectively, he points out, it was trading off simulation accuracy versus the time it took. "If you use enough detail to give you confidence in the results, it takes too long and you get to the point where it's not practical in the development timeframe. If you're looking for accurate outputs, you need accurate inputs to start with and that's a pretty complicated proposition."
Engineers were telling Reaction Design's team that real fuels were too complex to model accurately and they didn't have reliable models as to how the fuels would act in the cylinder. FORTÉ was hatched from a Reaction Design-led, industry-funded project called the Model Fuels Consortium (MFC). When starting MFC in 2005, Reaction Design discovered an absence of models of what fuel chemistry really looked like, so they assembled a consortium of suppliers and what Rosenthal describes as "demanders of energy" to develop accurate fuel model chemistry mechanisms for use in this simulation, whether that's diesel or petrol. "We started with seven or eight companies and that has now grown to over twenty, including VW and PSA, as well as a number of Japanese and American OEMS, and we're adding to the consortium on an annual basis," states Rosenthal. The idea behind Reaction Design's pioneering concept of surrogate fuels is that fuels are composed of classes of different chemical molecules and every structure has a different behaviour, as Rosenthal explains. "If you find a representation of a fuel molecule, you can represent the behaviour associated with the class of the molecule, which is the case across all fuels. Even if you look at the alternative fuels coming in, you can represent those as sub sets of what they really are." The team has also developed surrogate blend optimisation software that takes the octane or cetane rating as one of the inputs, allowing engineers to describe some of the basic formulations and take into account variations from one part of the world to another, or if it's a heavier or lighter mix. Also, they've started to add basic ethanols and second generation biofuels. "As they roll in, we're adding them to the database," he says. The problem is that, once you exceed 1,000 chemical species, it can takes weeks to get a solution for one engine cycle and that's even more impractical. "It can take upwards of 80% of the calculation of that engine cycle; that's an area which is being ignored, because engineers haven't got the time to get the answers. Under pressure and ignition, the flow changes, the temperature gradient changes, so we have tens of thousands of calculations that have to be included and project what the molecules are going through. "The complexity of both the mechanical and chemical process, and the way that, for example, when
you inject a spray how it is disseminated in the cylinder, has a bearing on all the factors you're interested in, such as the temperature, where the flame starts and how it propagates.
"There are a number of complex physical and chemical facts taking place in the engine during the cycle, so you have to be able to manage them all," he points out. Reaction Design, explains Rosenthal, "had to go back to the drawing board and break new ground on how those chemical equations were solved, to the point where we had to bring together some numericalsolving algorithms that haven't typically been used. "We were successful in grouping the complexity from an exponential relationship to a linear relationship.
This basically says that previously, when you went from 35 species up to 150, the time it would take to simulate that model would increase by nearly a factor of ten. To reduce complexity, we employ an 'on the fly' reduction mechanism that uses a smart algorithm that looks at the next cell simulation and decides whether the chemistry in that next cell is required for the solution. If it's not, we don't use it. Another thing is cell clustering where, if it has already calculated the chemistry for a particular cell and it's the same chemistry in an adjacent cell, it doesn't recalculate, but just reuses the answer."FORTÉ uses a mesh model for spray droplets that's intolerant of grid size, resulting in higher accuracy and, because it's a very coarse grid size, reducing the amount of calculation and subsequently simulation time. It also employs an auto-mesh generation feature that works around geometric boundaries, with local meshing around the system's edges or the injector nozzles to allow the preview of mesh upfront before the simulation. Unlike many auto-meshers that employ odd-shaped cells that are hard to populate mathematically and are computationally intensive, or a cut cell approach that can result in rounding errors, FORTÉ employs a fully rectangular or Cartesian cell with an immersed boundary approach that contains all the data and then eliminates that which isn't germane to the calculation, as opposed to other approaches where missing data is inferred. This, claims Rosenthal, has a "large" effect on both accuracy and reducing simulation times – from 133 to 13 hours. "The set-up is more straightforward, with a lot less time to calibrate to make sure the spray model is correct and you don't have to export to another set of tools. So overall there's a 25-50% improvement in the time to solution." Rosenthal claims FORTÉ is the only tool available today that can handle multiple injections of multiple fuels. "We're working with European and American vendors that are looking at mixing diesel with gasoline or natural gas and getting some very interesting results, including a dual gasoline-diesel engine that runs 35% more efficiently than the standard engines." And since FORTÉ can ead to the creation of engine designs that result in a cleaner, more complete burn, it could help minimise catalyst loadings. "I think we can get more detail and insight into how much we can eliminate in the cylinder before it gets out to an after-treatment and that's where a lot of the value of this is shown."
FORTÉ, he maintains, could help speed up the development and understanding of advanced engine technologies, such as pre-mixed charge compression ignition (PCCI), homogenous charge compression ignition (HCCI) and dual fuel systems, "It helps engineers to understand what's happening in HCCI to make the switch over smoother. Much of the work we see today is some form of charge compression ignition, HCCI or PCCI." Meanwhile, energy suppliers could also benefit from FORTÉ's capabilities to understand further howfuels are burnt, in order to deliver even cleaner and more efficient fuels to the end users. "We've been working with some of our partners for six to nine months, so we're still a couple of years out, but we're getting feedback that it's already altered design frames. "One of the things we've done is allow people to import experimental data and create views next to modelling data, a closed loop system – and take a look at projected versus what's being seen on the test bench and then modify the model's parameters or some of the environmental boundaries. That's the last piece of what engineers were asking for."
|
|
| |
Author
Ian Adcock
|
| |
| |
Copyright AD Media Europe.
See Terms and Conditions.
One-off usage is permitted but bulk copying is not.
For multiple copies contact the sales team.
|
| |
|
|
| |
To access the whitepaper section you need to complete our 60 second registration process. Once completed this then allows you to download any and all white papers, register for e-zines and access our detailed supplier directory for FREE.
If you are all ready a registered user then enter your e-mail address and login.
You will need to have logged in prior to entering your comments in the boxes provided.
|
|
Engineering Recruitment Partner:
|