Heat transfer coefficient in regenerator
 

Hello everyone, I'll show you my model for my thesis and I would like to analysis some details for the conjugate heat transfer (CHT), especially the heat transfer coefficient (h) and Nusselt number (Nu).
https://www.youtube.com/watch?v=8NADigv5LT8
So, I've got reversed fixed-matrix regenerator who has following assumptions and BC:
1)There are 2 domains - 1 solid domain - ceramic material and 1 fluid domain - air. I've got couple mesh interface for CHT, and there is heat transfer between solid and fluid domain;
2) The regenerator operates under transient flow or quasi-steady-state connditions (i.e., having constant mass flow rate and inlet temperature);
3) Heat loss to or heat gain from the surroundings is negligible so for BC my walls of the solid domain are adiabatic - the heat flux q = 0 W/m2;
4) The inlet velocity and temperature are uniform over the flow cross section and constant with time, therefore I use velocity inlet for BC;
5) The outlet BC I set pressure outlet;
6) The Initial conditions for the model are initial temperature of the solid domain so I patch the temperature in Solution Initialization;
7)I setting up k-epsilon turbulence model;
8) So I solved the model for 1000 iterations, number of time steps 100 and time step size 0.6 s;
9) After that when I've got this calculations I save my results, and I made duplicate for my model in workbench, because I have to calculate the reverse flow i.e. I use my calculation from last model to create new reverse model from Calculation activities - Automatically Initialize and modify case I open my last calculation file .dat.gz.;
10) For the reversed model my inlet become outlet and my outlet become inlet and the parameters are similar.
Everything looks good and similar with my experimental data in my prototype, but when I compare the heat transfer coefficient in Ansys Fluent and my experiment, the software give me absurd results. When I report in surface integrals -> wall fluxes I have two options to use "Surface heat transfer coefficient" or "Wall function heat transfer coefficient" for the both gives me some error from my experimental date. So I would like to know what is the difference between these coefficients how they are solved in Fluent.
I apologize for my English and I'll be glad somebody to reply of my problem, Thanks a lot! :)