在内燃机设计领域中，企业不仅需要在同行竞争中保持领先，而且还需积极探索，努力遵循新的排放法规要求。

计算流体动力学（CFD）工具可协助发动机设计师不断努力，设计出更高性能、更低排放的发动机，且期间无需打造成本高昂的实体原型。在CFD工具的协助下，设计师可以直观地观察、测试虚拟燃烧室中的模拟燃料燃烧与点火情况，打造更加轻松、清洁、高效的发动机。

显而易见，只有当模拟结果能够真实反映客观事实时，CFD技术才有其存在的价值。因此，为了准确预测发动机的真实性能与排放情况，设计师的模拟必须精准反映燃烧过程中的化学动力学变化。需要指出的是，经过简化的燃料化学模型，以及哪怕最精细的网格划分技术，均无法实现这一目标，而如果模拟手段不能发挥作用，设计师们就不得不重返成本昂贵且耗时久远的物理测试手段。

CFD技术应对“燃烧模拟挑战”

具体来说，利用CFD技术模拟燃烧过程非常复杂，需要进行大量的计算，特别是在一些烟炱形成或发动机震爆等过程中尤其如此。传统的多层CFD模拟需要设定几千个变量，计算过程可能需要数天才能完成。此外，为了追求最优设计，设计师通常会一点点调整发动机的几何尺寸和燃料种类，进行大量尝试，这个过程很有可能需要持续数周，甚至数月的时间。

为了加快设计速度，许多CFD解决方案均选择了简化的燃烧化学模型，并寄希望于异常精细的网格划分技术来弥补因此损失的精确度。然而，由于这些简化模型通常来源于一些未经缜密验证的第三方资源，因此很容易出现来源不一致或互不兼容的问题。在模拟过程中，由于不同来源的模型可能会重复反映某些物质或某些反应过程，因此会给燃料的配比与定制带来很大困难。

现如今，发动机燃油的配方经常随季节（例如，美国夏季汽油配方中的丁烷含量高于冬季配方）、地区（例如，美国地区柴油的属性与欧洲柴油不同）及具体应用场景的不同而变化。此外，乙醇和生物柴油等替代燃料也已成为化石燃料的有力补充。在此背景之下，设计师必须借助更强大的模拟工具才能处理成分日益复杂的发动机燃料。

为了探究这部分燃料的性能，我们必须确保模拟中采用的燃料模型能够反映真实的化学过程。但遗憾的是，传统CFD软件包中自带的燃料模型都采用了非常复杂的算法，且计算时间很容易成倍增加，特别是在涉及多成分燃料的情况下。

对烟炱形成和发动机爆震进行模拟

目前，最新的颗粒物（PM）排放法规已经对发动机设计师提出了特别挑战。众所周知，由于物理化学反应过程复杂，传统的CFD工具很难模拟烟炱的形成过程。因此，在传统的内燃发动机设计中，设计师为了改善烟炱的生成情况，通常需要进行为期多年的原型制造与测试。

发动机震爆是指燃烧室内的高压燃料与空气混合物的自燃，通常发生在驾驶员点火之前或之后的时段。为了准确预测发动机的震爆，设计师必须准确模拟火焰头在进入燃烧室时的位置与结构。不过，传统的CFD工具主要基于简化的化学过程与精细的网格划分技术，因此通常很难准确模拟自燃的过程。

由于火焰头的厚度远远小于模拟工具中的网格尺寸（哪怕是最精密的网格划分），为了确定火焰头的位置，传统CFD模拟工具需要借助数量非常庞大的网格才能解决火焰拓扑的难题。然而，由于需要借助的网格单元数量过于庞大，为了保证稳定性，这种CFD模拟通常需要划分成极细的步骤分步进行，因此耗时非常漫长。

作为发动机设计者需要怎么做呢？

从化学动力学入手

燃烧的核心就是化学反应。因此，为了准确预测发动机的真实性能表现，CFD模拟必须抓住真实化学反应的核心--化学动力学。

举例而言，ANSYS公司的内燃发动机模拟工具ANSYS Forte并未采用公认的尺度公式，而是转而采用时间尺度。这样一来，设计师可以根据真实需求尽可能多地加入各种反应，整个过程并不会增加模拟时间。因此，即使一些规模更大且准确度要求更高的燃料模型, 也能达到与“简化版”模型相仿的计算速度。在这种情况下，设计人员可以快速、准确地预测发动机的排放情况，且大幅减少实体硬件原型的使用需求。

ANSYS Forte工具支持行业标准版燃料模型库（Model Fuel Library），可直接使用其中的大量燃料模型，其中包括很多新型混合燃料模型。

在精确化学燃料模型的协助下，设计师得以大幅提升燃烧过程的模拟质量，进而更加敏捷、高效地应对日益严苛的监管条例要求，并打造真正先进的清洁发动机与燃料技术。

For internal-combustion (IC) engine design, companies need to stay ahead of not only the competition but also of new emissions mandates.

Computational fluid dynamics (CFD) can help engine designers create higher performance, lower emissions IC engines without costly physical prototyping. CFD lets engine designers visualize and test fuel and ignition behaviors within a virtual combustion chamber, providing a faster way to design cleaner, more efficient engines by simulating ignition and fuel dynamics.

But combustion CFD can only be of value if the results predict real-life behaviors. To predict actual performance and pollutant emissions, simulations need to accurately account for the chemical kinetics of the combustion process. Simulations that rely on drastically simplified fuel chemistries and ever-finer meshing technologies fall far short of this goal. And when they fall short, designers must fall back on expensive, time-consuming physical testing for answers.

Combustion CFD challenges

Combustion CFD is complex and computation-intensive, especially for applications such as soot formation and engine knock. Times can easily stretch into days for traditional CFD multi-stage simulations with thousands of variables. Incremental changes to engine geometries and fuel models can stretch the total time to weeks or months before an optimized design is realized.

To speed design time, many CFD solutions simplify combustion chemistry, trusting that severe mesh refinements can make up in detail what they lack in precise chemistry. These simplified fuel models rely on weakly validated, third-party mechanisms from disparate and incompatible sources. Using models from multiple sources makes it very difficult to blend or customize fuels in simulations because species and reactions may be duplicated—perhaps in a contradictory way—in different sources.

Better models are needed now because motor fuels have become more complex. Fuels vary by seasonal formulation (U.S. summer gasoline contains less butane than winter formulations), by region and by application (U.S. diesel has different properties than European diesel). Alternative fuels such as ethanol and biodiesel now supplement petroleum-derived fuels.

To understand the effects of these diverse fuel types, chemically correct fuel models are required. Unfortunately, fuel model algorithms in conventional CFD packages are complex and compute time can multiply exponentially when they are combined to represent multicomponent fuels.

Simulating soot formation and engine knock

New particulate matter (PM) regulations present particular challenges for engine designers. Soot phenomena are notoriously difficult to simulate and too complex to run in conventional CFD software, due to the physics and chemical reactions leading up to soot formation. As a result, optimization of soot in conventional combustion-engine design typically requires years of building and testing prototypes.

Knocking occurs when the highly compressed fuel and air mixture in the combustion chamber auto-ignites, either before or after the spark that is meant to trigger ignition. Accurately modeling the location and structure of the flame front as it expands into the combustion chamber is extremely important for predicting knock. But simulating auto-ignition is very difficult with conventional CFD approaches that rely on mesh refinement and simplified chemistry.

Since the scale of the flame front thickness is significantly smaller than computational mesh—even with severe grid refinement—CFD simulations that rely on mesh to resolve the flame location will require an inordinately large number of tiny cells to resolve the flame topology sufficiently. Simulations with large numbers of small cells can easily get bogged down by the tiny time steps needed to maintain simulation stability, and require an impractical amount of computation time.

What’s an engine designer to do?

Follow the chemistry

Chemistry is crucial. It’s at the heart of combustion, and for internal-combustion CFD to accurately predict real-world engine behavior, it must precisely account for real chemical kinetics.

ANSYS Forte, for example, changes the well-established scaling equation: instead of scaling with the cube of the number of species, the simulation time scales linearly. This allows an engine designer to include as many reactions as he or she requires for accurate simulations, without incurring a compute time penalty. Even larger, more accurate fuel models achieve compute times comparable to those with severely reduced, less accurate models. As a result, designers can quickly and accurately predict emissions that translate reliably to actual engine designs—with far less trial-and-error hardware prototyping.

Fuel components derived from the industry-validated Model Fuel Library enable ANSYS Forte to simulate combustion for a large variety of new or existing fuel blends and foresee what emissions will occur for a wide range of operating conditions.

By using accurate fuel models based on precise chemistry, engine designers greatly increase the predictive quality of combustion simulations, to more quickly and effectively meet strict regulatory guidelines and create advanced clean engine and fuel technologies.

Author: Bill Kulp
Source: SAE Truck & Off-highway Engineering Magazine