Hello everybody,

I am working on a pipeline-project. The problem i am facing is the following: during winter the external walls of the pipelines are frozen. Sending hot fluid through the pipeline will have an impact on this external ice-layer (in what way ever). Does anybody know a commercial cfd-sofware that is able to handle the internal-flow problem and the impact on the ice-layer at the same time ? any help is very appreciated !

At Opel they did some work for de-icing the windshield of a car resulting from the internal air-flow. They used StarCD with some additional user coding. So you might ask the guys from Computational Dynamics in Nuernberg.

CFD-ACE+ from CFD Research Corporation has been used to study de-icing on the windshield of a car. Melting of ice on the external part of windshield, which resulted from the internal HVAC flow, was predicted in a transient simulation. This example is also posted on our website http://www.cfdrc.com under the automotive applications. No user-subroutine is needed to use CFD-ACE+ for this application.

Based on the information given, without knowing about other complications, it seems to me that you can solve this problem analytically or at worst using a simple numerical scheme to solve the system of equations that describe this heat transfer problem. Using CFD appears to be an unnecessary expensive proposition.

Adrin Gharakhani

Is the pipeline exposed to air or buried? It may not be so easy as Adrin imply.

Andrzej

the pipeline is exposed to air. since i am trying to optimice the airflow inside the pipeline under consideration of heat transfer to the ice layer and therefore variable ice layer size (due to melting) i agree with your statment: it might not be that easy !

I actually assumed the pipe was exposed to air when I suggested you don't need CFD. I can understand that for a car windshield case you will need CFD because the problem is inherently 2-3D, but the pipe is easier.

So you have an axisymmetric pipe with its walls and possible insulation and then an ice layer with an initial thickness (which decreases in time) and then air flow.

This is a quasi-1D unsteady problem that can be solved iteratively and quite accurately without CFD. There are excellent analytic-empirical formula that relate heat transfer and flow properties (Nu vs. Re) for flow inside a pipe as well as flow outside the pipe. These are generally based on boundary layer theory which for the simple case of flow in a pipe (no separation) are as accurate as any CFD analysis.

So you stack these layers of outside air, ice, insulation, pipe wall, inside air, and apply energy balance to the stacks (keeping in mind that ice thickness is variable).

I think you may be able to find the solution to similar problems in graduate level heat transfer books. I remember the same problem (with the added complexity of a vertically-inclined - not horizontal - pipe) given as homework to an undergrad heat transfer class at MIT.

Adrin Gharakhani

(1).Excellant idea. (2). At least, it can be used as a reference solution. (3). For those who don't have a MIT degree, CFD solution is less painful to accept. (4). On CFD side, it can be done in the similar way in 2-D axisymmetric formulation or a slice of 3-D pipe, with internal fluid flow and external layers of ... ice...air flow. (5). I have to say that most CFD users in application field simply don't have the analytical capability to solve a real problem. And with most CFD codes are being used as a black box, the user is not getting the opportunity to learn the basic, I mean, to think, to formulate the equations, .....etc. "selecting" and "setting" some objects and conditions are basically what the users in CFD applications do to get the job done. "user-friendly" options are replacing the user's "ability to think". (6). But, I think, the problem can be solved in either ways. Regardless of the approach, various assumptions have to be made for this problem, in order to make it a simple problem.

> 6). But, I think, the problem can be solved in either ways. Regardless of the approach, various assumptions have to be made for this problem, in order to make it a simple problem.

Right on the money!

The "various assumptions ..." is exactly the point in not necessarily using CFD for this problem. Sure, under severe conditions of direct numerical simulation one would end up not getting a fully developed (and symmetric) temperature profile in the pipe because presumably the melted ice would turn into drops and streaks of water (NOT uniformly distributed water) which would roll down and around the pipe, thus further affecting heat transfer properties and complicating the problem.

But we know, short of a method that uses Volume of Fluid (or something similar) and all sorts of other gizmos, the average CFD package would force the user to make hard choices and assumptions. And I'd bet these will be the same assumptions that would be sufficient to get semi-analytic results!

Adrin Gharakhani

The problem is so simple only at the level of graduate courses. In real life it is not. As you mention the water will melt and trickle down destroying symmetry of the problem. Depending on the initial ice thickness one day may not be enough and in night it will refreeze in completely different form. One part of the pipeline can be exposed to sun and will melt much faster. The other part may freeze at the same time! We are far away from academic exercises, even though they may be a good point of reference.

Andrzej

Adrin is right the proposition that a big commercial cfd code is needed to solve this problem is ridiculous. i went to the University of Miami (not the great MIT, although my heat transfer professor did) and i could solve this problem quite easily with a few handbook heat transfer correlations and a simple matlab program. since many CFD codes don't give great heat transfer predictions using CFD might give poor answers anyway. all you have to do is calculate the mass flow and temperature thru the pipe that will provide enough heat transfer to melt the existing ice and any that may form overnite. the various effects you described (water trickling) does not affect the symmetry because you'll be doing a 1-d problem. as for the others (uneven ice build up etc) you overengineer the problem slightly to take variations into account. you'd have to do this with the cfd solution anyway unless you wanted to do an analysis for every possible contingency. if an engineer who uses cfd can't do this simple analysis God help him with the CFD.

I have even simpler solution. Wait until July and ice will disappear for sure. Then the analytical solutions will be just irrelevant. Actually, they are irrelevant for this problem right now.

I agree in one point. Traditional CFD software was not design to deal with objects that are exposed to natural environment.

May be people that want to spend private money for simulation of pipeline melting/freezing are not so stupid as it might have appeared from the point of view of undergraduate exercises.

Andrzej

(1). I just want to point out that I am not so sure the ice will disappear in July. Sometimes, winter comes in July. (2). The problem can be simplified to 1-D . And for the non-symmetric heating problem, it can be treated as a 2-D problem. (3).If one has to cover several hundred miles of pipelines, 3-D is probably not feasible. (good for car wind shield icing problem, 3-D CFD is all right.) (4). The other problem is that the boundary conditions are difficult to fix in the real environment. In this case, some simple 1-D simulations will bring out the key parameters for this problem. And if one is still interested in the sun's position, then a 2-D simulation can be used to answer the sysmetry question.

Hi 'Wilhelm',

what about heating the entire pipeline? ;-)

By the way: What kind of fluid is pipeline for?

Thomas.