How to calculate entropy generation in (W/mK) in CFD-Post?
 

Dear all,

Hello. Hope all are well. My question is simple. How can I calculate entropy or entropy generation in CFD-Post for flow (incompressible) in pipe from fluid flow and heat transfer fields (i.e. solving momentum and energy equations)? I didn't find much in ANSYS documentation except a formula "s-s_ref=c_p*log*(T/T_ref)" for constant density and specific heat flow. Would I be making an expression or anything else? Also, I know static entropy variable cannot be used as the fluid is flowing and not static (and even in compressible flow for isentropic conditions there is not distinction between entropy and static entropy). Would be grateful for guidance. Thanks.

Based on the formula shown, you can define your entropy field (set an arbitrary s_ref and T_ref), evaluate

Then, compute some kind of average at the inlet and outlet. Say

EntropyChange = massFlowAve(MyEntropyField)@Outlet - massFlowAve(MyEntropyField)@Inlet

In ANSYS CFX, Static Entropy is just entropy (just a name).; therefore, you can just

EntropyChange = massFlowAve(Static Entropy)@Outlet - massFlowAve(Static Entropy)@Inlet.

Keep in mind if the flow is isothermal, the entropy change is 0 by definition of an incompressible fluid.

Thanks Opaque. What value should I use for s_ref and T_ref? Also what is "MyEntropyField" and where do I get it from? Moreover, the problem with static entropy is that it gave me units of m^2s^-2K^-1 whereas units of entropy generation per unit volume is W/m^3K. Why is that? Thanks.

Several questions.. Let us take one at a time.

For s_ref, T_ref, they are arbitrary values. For simplicity, just make them 0 [J kg^-1 K^-1], and 1 [K]

In CFD-Post, you can create your own variables and use expressions to evaluate them. Create one named MyEntropyField and go from there. Please read the documentation for further functionality.

Dimensions of specific entropy (or static entropy) are energy per unit mass per unit temperature. In SI, that is [ J kg^-1 K^-1] that expanded on basic units is [m^2 s^-2 K^-1]

Entropy generation per unit volume is mass flow times the change of specific entropy in the system. You can just replace the massFlowAve( ) operation by massFlowInt( ) operation and you should be good.

Thanks a lot :)

Dear all,

Hello. Hope all are well. I have used two methods to calculate entropy generation. First the Bejan correlation

(((Wall Heat Flux[W/m^2])^2*3.142*(D[m])^2)/(Thermal Conductivity[W m^-1 K^-1]*(Average T)^2*Average Nu))+((32*(Average MFR)^3*Average FFF)/(3.142^2*(Density[kg/m^3])^2*Average T*(D[m])^5))

where D is the diameter
Average T is the average temperature,
Average Nu is the average Nu number,
Average MFR is the average mass flow rate, and
Average FFF is the average Fanning friction factor

is giving me a value of around 0.3 Wm^-1K^-1

Second I have defined MyEntropyField
s_ref=c_p*log*(T/T_ref)

where s_ref is 0 [J kg^-1 K^-1] and (checked from CFX-Pre)
T_ref is 298 K (checked from CFX-Pre)

Applying
massFlowInt(Entropy Field)@Outlet - massFlowInt(Entropy Field)@Inlet

This gives me a value of -0.55626 Wm^-1K^-1

The two values are entirely different. I think the first one is more realistic as its positive and less than 1. Entropy shouldn't be negative, right? As the entropy of a system always increases and thus comes the concept of entropy creation. My question is what method should I follow and what can be the reason of such a big discrepancy? Would be grateful if some could help me out. Thanks.

Entropy is relative to a reference value. So yes, entropy can be negative.

You are comparing two different values as the averaging is different in both cases. For the Bejan correlation you average the input variables, but for the massflowInt() calculation you average the entropy field.

Dear all,

Hope all are well. I know this has become a relatively old post but as my question is related to the topic I thought to post it here. I want to find volumetric entropy generated as a continuous variable in units of [J K^-1 m^-3] an not the entropy generation rate in [W m^-1 K^-1]. Currently, I am using Bejan correlation and a defined entropy field. Would be grateful for help on this matter. Thanks.

Got it by defining gradients :) But now that I have got a numerically calculated variable entropy field, the question still remains to find it in units of J m^-3 K^-1. Can anyone please help me. Thanks.

Would really appreciate some help from experienced professionals. Glenn and Opaque maybe you guys have come across this question. Thanks.