While basic single-phase fluid flow modeling is now almost a commoditized service, it is still very difficult to complete projects that are even a bit more complicated.Â An example is two-phase flow with flashing, evaporation or boiling.Â Here we look at 2 different systems with unexpected corrosion patterns.Â The systems have different configurations, and we look at how this leads to very different ways to use CFD to model the system.Â
A wash water system simply sprays water into hot vapour going through the large-diameter pipe to ensure any corrosive components of the vapour are well diluted.Â Some of the water will evaporate from droplets in the spray, some willÂ hit the walls and form a surface film, some will evaporate from the film, and some will be re-dispersed into droplets from the film.Â If there is enough water that doesn’t flash, some may flow as a free stream on the bottom of the pipe and will take a path determined by gravity, vapour velocity and momentum.Â Heat is transferred to/from the pipe wall to the vapour and film water, and from the vapour to the water, including the heat of vapourization.Â That is a lot to model.Â Knowing that all computer models are limited in some ways, the first priority is to ensure the modeling effort doesn’t have limitations that invalidate the result, yet is still tractable and does not waste resources on modelling things that are less important to the question.
For the first system we consider, the overhead piping has a horizontal run followed by 4 vertical legs.Â It has large droplet sizes and lots of water.Â The first and last legs have higher corrosion rates.Â Why?Â Given that the amount of water is relatively large (i.e., compared to industry averages), and the droplet size was expected to be large, I thought drying surface film was less likely to be an issue than simple liquid distribution.Â Most of the water remains liquid and would follow the liquid path determined by the 3D shape of the overhead, gravity, vapour velocity and the fluid’s momentum.Â This model probably doesn’t need the detailed phase change modeling, at least for the first pass.Â Because we expect a sharp interface between water and vapour, and don’t care too much about small-scale details, we can use a VOF model with interface sharpening.Â We need to get viscosities and densities correct, but omit heat and phase change.Â It is modeled in 3D, but at a coarse scale (600K cells) for speed.
The video below shows how a wash water system distributes in a non-symmetrical overhead.Â The redÂ color represents about 6% water by volume, so the wall areas near the red would be wetted by water, while the areas without water droplets dispersed in the vapour phase are more likely to be dry.Â This simple model answers the key question “Why are some legs well protected, while others are not?”Â Clearly, the asymmetry is critical, and the changing vapour flow rates and water momentum affect the flow.Â The first leg gets little water because the high velocity of the coarse droplets allows momentum to carry the droplets past it.Â The last leg gets little water because the vapor velocity (with 1/4 of the flow) is too low to keep many large droplets suspended.Â Â
After this, there may be a desire to increase resolution and add other parameters, but since the question is already effectively answered, the expense is harder to justify.
Next, we will look at a more complicated system, where flashing is an issue.