Plate heat exchanger with pressure drop and heat transfer
 

Hello,

I'm trying to model a plate heat exchanger using CFX. The main idea is to model two channels, giving one interface, and then by using symmetry conditions simulate that these two channels are in the middle of the heat exchanger.

Furthermore, to keep the geometry as simple as possible the channels are represented by a rectangular box as the heating area, and by two triangles in each end as the distribution areas.

In reality the channels are separated by a thin corrugated metal plate, but since it's too complicated to draw this pattern the plate is modeled as a plain wall instead. To compensate for this and accurately represent the pressure drop I'm using some empirical correlations relating the the Reynolds number to the Fanning friction factor (f=c1*Re^c2). The plan is to use this correlations to simulate the flow profile and the pressure drop. This is done by adding a subdomain and under Sources -> Momentum Source/Porous Loss introduce a General Momentum Source. The X Y Z components have the units of [kg m^-2 s^-2], so I multiplied the friction factor like this: (4*Fanning*1/hydraulic diameter*rho*velocity^2/2). I assume that CFX then calculates this resistance over the length of the node, giving the unit [kg m^-1 s^-2 = Pa]. Is this correct? What do you think of this approach to simulate the pressure drop and the flow profile?

My current model consists of two fluid domains in the shape of the two channels. The both domains are identical, the only thing that is different is the inflow temperatures and the flow directions. The domain with the hot channel is working perfectly fine, I get nice streamlines and a reasonable pressure drop of about 50 kPa. However, the cold channel is not converging at all. The solver keeps giving me the following message: "A wall has been placed at portion(s) of an OUTLET boundary condition (at XX.X% of the faces, XX.X% of the area) to prevent fluid from flowing into the domain."

This problem only occurs for the cold channel (inlet temp 25 deg C) but never for the hot channel (inlet temp 60 deg C). For both channels the outlet boundary is set to average static pressure and 0 Pa. I've tried with the alternatives with using an opening instead, changing the outlet condition to mass flow rate out and normal speed, but nothing works. I don't want to make an extension of the channel. Besides, the cold channel is identical to the hot, so I don't see how one of them could work but not the other. Any suggestions?

Thanks in advance! :)

Add a picture for explanation of you setup.

The "a wall has been placed..." message is an FAQ. See https://www.cfd-online.com/Wiki/Ansy...f_an_OUTLET.22

Is this one of those brazed plate heat exchangers? Just so you know, the insides of those are very complex. I don't think the heat transfer will be represented well at all by just a flat plate, it will be highly under-predicted.

Like I said in my original message, the alternatives listed in that FAQ didn't help/wasn't an option and that's why I asked for a different solution. The problem was partly solved by replacing one mutual parameter for the General momentum source in X and Y direction with separate parameters, one for X and one for Y.

Well, I thought it to be gasketed and not brazed, but the inside should probably look the same. I know that the inside is very complex, and that's the whole reason for this simulation, to see if it can be represented by the plain plate instead. I have some correlations for the heat transfer coefficients that describes the heat transfer for the pattern at different flow rates.

In order to keep it simple and fast I have two water domains but no steel domain separating them. Instead I have chosen the interface to be Side Dependent. Thoughts about this?

So what, you will just increase the conductivity of the fluid to add to the heat transfer and make it the same as your correlations?
What will you be figuring out with the simulation then if you already have the correlations? Just use the normal hand calcs for heat exchangers and the LMTD method. That will be more accurate than the CFD and much much simpler.