The Nusselt Problem​
​Authors: Housam Binous and Ahmed Bellagi

We consider the problem of heat transfer in plug flow heat exchanger .
​
This problem is similar to the Graetz problem where the velocity profile is laminar .
​
Axial conduction is ignored .
​
We determine the temperature profile obtained using NDSolve at different local residence times .
​
We also provide a solution based on orthogonal collocation.
​
Description of the orthogonal collocation method using Chebyshev-Gauss-Lobatto points is given in the following paper:
​H. Binous, A. Shaikh and A. Bellagi, Chebyshev Orthogonal Collocation Technique to Solve Transport Phenomena Problems With Matlab and Mathematica, Computer Applications in Engineering Education, 23(3), pp. 422–431, 2015.​
​
We show that the analytical solution (derived in details in Ref. 1) gives the same result as their counterparts obtained using NDSolve and orthogonal collocation.
Acknowledgement :​​Governing equations and analytical solution of the Nusselt problem are reproduced from Ref. 1.​​
​​In dimensionless form, the problem reduces to :​​
​​where​​
​​​References:​​1- Rice and Do, Applied Mathematics and Modeling for Chemical Engineers, 2nd Ed. , New-York: Wiley, 2012, pp. 252-253.​
In[]:=
Off[General::"spell1"]
Solution using NDSolve
In[]:=
sol=NDSolveD[θ[τ,ξ],{τ,1}]
1
ξ
D[ξD[θ[τ,ξ],{ξ,1}],{ξ,1}],​​θ[0,ξ]Exp[-1000(1-ξ)],θ[τ,1]1,(D[θ[τ,ξ],{ξ,1}]/.ξ10^-8)0,​​{θ[τ,ξ]},{τ,0,10},{ξ,10^-8,1}
Out[]=
θ[τ,ξ]InterpolatingFunction
Domain: {0.,10.},1.×
-8
10
,1.
Output: scalar
[τ,ξ]
In[]:=
plt=Plot[{First[θ[τ,ξ]/.sol]/.τ->1,First[θ[τ,ξ]/.sol]/.τ->0.1,First[θ[τ,ξ]/.sol]/.τ->0.01},{ξ,10^-8,1},PlotRangeAll,PlotStyle{Darker@Red,Darker@Blue,Darker@Green},Frame->True,GridLines->Automatic,​​FrameLabel->{Style["dimensionless radial position",14],Style["dimensionless temperature",14]},ImageSize->400{1,1},AspectRatio->1,PlotLegends->{Style["τ=1",14,Darker@Red],Style["τ=0.1",14,Darker@Blue],Style["τ=0.01",14,Darker@Green]}];
In[]:=
Show[plt]
Out[]=
τ=1
τ=0.1
τ=0.01
Analytical Solution gives same result as the one obtained using NDSolve​​
​​
​​
In[]:=
n=1;
In[]:=
While[n<10,{
λ
n
=FindRoot[BesselJ[0,x]0,{x,nPi}][[1,2]],Print[
λ
n
],n++}]
2.40483
5.52008
8.65373
11.7915
14.9309
18.0711
21.2116
24.3525
27.4935
In[]:=
θanal[τ_,ξ_]:=1-2
9
∑
n=1
BesselJ[0,
λ
n
ξ]
λ
n
BesselJ[1,
λ
n
]
Exp[-
λ
n
^2τ]
In[]:=
τ=1;
In[]:=
plt4=Quiet@Plot[First[θanal[τ,ξ]/.sol],{ξ,10^-8,1},PlotRangeAll,PlotStyle{Dashed,LightGray,Thick}];
In[]:=
τ=0.1;
In[]:=
plt5=Plot[First[θanal[τ,ξ]/.sol],{ξ,10^-8,1},PlotRangeAll,PlotStyle->{Dashed,Gray,Thick}];
In[]:=
τ=0.01;
In[]:=
plt6=Plot[First[θanal[τ,ξ]/.sol],{ξ,10^-8,1},PlotRangeAll,PlotStyle{Dashed,Gray,Thick}];
Comparison results with NDSolve and analytical solution
In[]:=
Show[plt,plt4,plt5,plt6]
Out[]=
τ=1
τ=0.1
τ=0.01
Solution using orthogonal Collocation and Chebyshev-Gauss-Lobatto points
Comparison results with NDSolve (solid lines) and orthogonal collocation solution (dots)