Specific Dissipation Heat Exchanger (TL-TL) heat transfer variation for different temperature levels

Hello together,

I am modelling a heat exchanger for an e-drive cooling system using the “Specific Dissipation Heat Exchanger (TL-TL)” component. While comparing the Simscape results to data from CFD simulations, I noticed that there is a deviation between results for the reference temperature at which the SD(mdot1, mdot2) correlation was specified and other temperature levels. The physical reason behind the change of heat transfer coefficients over temperature is that for changing fluid temperatures, also the fluid properties, like specific heat or dynamic viscosity, change. Therefore, the heat transfer coefficient of the heat exchanger varies. Is there a way to simulate the heat transfer (SD) dependent on mass flow rates and fluid temperatures? That would be very useful, as I have data for -40°C to 120°C available to specify everything dependent on both, mass flow rates and temperatures.

For the pressure drop calculation I noticed the same issue. It is specified using only the associated mass flow rate. I think there is some extrapolation for other temperatures going on in the background based on the reference values, but still the results are not satisfying. Especially for colder temperatures. To mitigate this issue, the 1D-simulation software AMESim uses the fluid temperature as a second input axis. The result is that the pressure drop data generated with AMESim matches the reference data also for fluid temperatures that are further away than ±20K from reference conditions. Here, kind of the same question as above occurs: Is there a way to compute the pressure correlations directly dependent on the fluid temperature?

Thank you in advance for your answer!

[Yifeng Tang]

Hi @Emil,

For the heat transfer part, it sounds like you want to implement a 4D table, SD(mdot1,mdot2,Tin1,Tin2), is that right? And you have data to produce such a table? This is not that far from implementing a heat flow directly, I would say.

I wonder if you calculate SD as SD = Q / (Tin1-Tin2), and plot a 3D scatter as SD(mdot1,mdot2), what does it look like? If it does look like a "volume", yes, there is a strong dependency on temperature (which I need to think why).

[Emil]

Hi @Yifeng Tang, that is right. And indeed, the current state of my model uses the available data for a direct calculation of the heat flow.

Regarding the 3D plot of the heat transfer coefficients: Yes, the graph looks like a volume. Here is a screenshot of it for one temperature level:

The SD values are on the z-axis, oil mass flow is on the x-axis, and coolant mass flow is on the y-axis. The values are from a constant fluid temperature level on both sides of the heat exchanger. If I add the data of another fluid temperature level, I get a second plot that has a similar shape but an offset in the vertical direction. For higher fluid temperatures the new graph hovers over the reference graph. For lower temperatures it is below the reference graph.

I think this behavior is comprehensible if we have a look at the NTU method for the calculation of heat exchangers. There is a relation between the heat transferred in total and the heat capacity flow of the fluids. Simplified, the heat transfer goes up if the available heat capacity flow is higher. The specific heat of both fluids, coolant and water, is proportional to the temperature. If the temperature level goes up, the heat capacity flow rate does so too. As a result, the Q rises. If we divide it by the temperature difference, we get a higher SD value as before, because the difference between the two inlet temperatures stayed the same. Only the overall temperature level was changed.

Furthermore, the viscosity changes with temperature. This change is especially distinct for low-temperature operation below 0°C. The viscosity directly influences the Reynolds number, which describes the flow characteristic of the fluid inside the heat exchanger. The flow characteristic itself influences the heat transfer coefficient between the fluid and the metal (laminar/turbulent). That's why the SD value changes with fluid viscosity.

[Yifeng Tang]

Hi @Emil,

Thanks for the details. It does suggest that the SD as a function of flow rates is not sufficient for your case. Directly driving the heat flow using lookup table and heat flow source will work fine, but you also need to be careful about near-zero or near-reverse flow conditions and don't add heat when there is almost no flow.

One thing you can try hopefully quickly is to parametrize a System-Level Heat Exchanger (TL-TL), using a condition of intermediate level of flow, temperature, heat flow, etc. as the "nominal condition" this block asks for. Check how it performs in other conditions with default correlation values. It won't allow a table as input, but the Colburn coefficients are tunable if the results deviate for other flow & temperature conditions.

[Emil]

Hi @Yifeng Tang,

thank you for the hint. I will try the System-Level Heat Exchanger too.