Select basis functions from the eigenvalue problem defined by the equipment geometry and its boundary conditions. For rectangular pilot plant domains, use sine and cosine eigenfunctions, usually as products across the Cartesian coordinates. For cylindrical domains, use Bessel functions in the radial direction, Fourier sine or cosine functions in the angular direction, and suitable trigonometric functions along the axis. The correct functions are the ones that satisfy the homogeneous form of the physical boundary conditions, because they then form an orthogonal basis for calculating expansion coefficients.

The geometry determines the family of eigenfunctions, while the boundary conditions determine the allowable eigenvalues and the specific sine, cosine, or Bessel combinations. Basis functions that match both produce independent, orthogonal modes and make diffusion or conduction models substantially easier to solve.

Start With the Spatial Eigenvalue Problem

Separate the governing equation by coordinate system

Transient conduction and diffusion models commonly reduce, after separation of variables, to a spatial eigenvalue problem such as

[ \mathcal{L}_x\Phi=-\lambda^2\Phi, ]

where (\mathcal{L}_x) is the spatial part of the heat or mass transfer operator.

The appropriate basis functions are the eigenfunctions of this operator under the pilot plant's boundary conditions. They are not selected solely because they are mathematically convenient.

Match the basis to homogeneous boundary conditions

A boundary condition may prescribe the value of temperature or concentration, prescribe its flux, or relate flux to the surface value.

These correspond broadly to:

Dirichlet: (\Phi=0), representing a fixed value after any steady component has been removed.
Neumann: (\Phi'=0), representing an insulated or zero-flux boundary.
Robin: (\Phi'+A\Phi=0), representing convective or interfacial transfer.

For nonhomogeneous conditions, the field is often decomposed into a steady or lifting solution plus a transient remainder. The eigenfunctions are then chosen to satisfy the homogeneous boundary conditions of that remainder.

Selecting Functions for Rectangular Geometries

Use sine and cosine eigenfunctions in Cartesian directions

For a one-dimensional rectangular region, the spatial equation has the form

[ \frac{d^2\Phi}{dx^2}=-\lambda^2\Phi, ]

with general solution

[ \Phi(x)=a\sin(\lambda x)+b\cos(\lambda x). ]

The boundary conditions determine which combination is allowed and which discrete values of (\lambda) satisfy both boundaries.

Use products for multidimensional rectangular equipment

For a rectangular slab, channel, catalyst section, or flat-plate component, separated modes typically have the form

[ \Phi_{mn}(x,y)=X_m(x)Y_n(y), ]

or, in three dimensions,

[ \Phi_{mnp}(x,y,z)=X_m(x)Y_n(y)Z_p(z). ]

Each factor is selected independently from the boundary conditions on its corresponding pair of surfaces. The full mode is orthogonal because the one-dimensional factors are orthogonal.

Apply the boundary condition to choose sine or cosine modes

A zero-value boundary at (x=0) eliminates the cosine term, producing sine modes. A zero-flux boundary at (x=0) eliminates the sine term, producing cosine modes.

For example, on an interval (0\leq x\leq L):

(\Phi(0)=\Phi(L)=0) leads to sine modes such as (\sin(n\pi x/L)).
(\Phi'(0)=\Phi'(L)=0) leads to cosine modes such as (\cos(n\pi x/L)), including a constant zero-eigenvalue mode.
A fixed condition at one end and a zero-flux condition at the other produces a shifted eigenvalue sequence, commonly involving half-integer multiples of (\pi/L).
Convective boundaries produce Robin eigenvalues determined by a characteristic equation rather than by the simplest integer multiples.

The exact eigenvalues depend on the dimensions and boundary parameters of the equipment.

Selecting Functions for Cylindrical Geometries

Use Bessel functions for radial variation

In cylindrical coordinates, the radial part of the diffusion or conduction operator contains the factor (1/r). The radial eigenvalue equation therefore differs from the Cartesian equation and leads to Bessel functions.

For axisymmetric problems, the radial solution is commonly based on

[ J_0(\lambda r), ]

where (J_0) is the Bessel function of the first kind of order zero. Depending on the domain and boundary conditions, other combinations or orders may be required.

A solid cylinder that includes the axis normally excludes the singular Bessel solution (Y_\nu(\lambda r)), because it is not finite at (r=0). An annular region, by contrast, may require a linear combination of (J_\nu) and (Y_\nu), since the domain does not include the axis.

Add angular Fourier functions when the process is not axisymmetric

If temperature or concentration varies with the angular coordinate, the angular eigenfunctions are

[ \cos(m\theta) \quad\text{and}\quad \sin(m\theta), ]

with integer azimuthal mode number (m).

For a fully axisymmetric tubular reactor, shell, or column model, only the (m=0) mode is retained, so the angular dependence disappears. Non-axisymmetric heating, wall conditions, or feed distribution requires the higher angular modes.

Use trigonometric functions along the cylinder axis

For a finite cylinder or tubular pilot unit, the axial coordinate generally produces sine and cosine eigenfunctions, selected according to the inlet, outlet, end-wall, or symmetry conditions.

A separated cylindrical mode can therefore have the structure

[ \Phi_{mnp}(r,\theta,z)

R_{mn}(r), \Theta_m(\theta), Z_p(z), ]

where (R_{mn}) is a Bessel-based radial function, (\Theta_m) is a sine or cosine angular function, and (Z_p) is an axial trigonometric function.

Let Boundary Conditions Determine the Eigenvalues

Dirichlet conditions select zeros of the eigenfunction

If a cylindrical wall has a prescribed temperature or concentration, the radial eigenfunction must vanish at that wall after homogenization.

For a solid cylinder with (\Phi(R)=0), the radial eigenvalues are related to the zeros of the relevant Bessel function. This is the cylindrical counterpart of the sine eigenvalues obtained from a rectangular domain with two fixed-value boundaries.

Neumann conditions select zeros of the derivative

An insulated cylindrical wall requires zero radial derivative:

[ \left.\frac{d\Phi}{dr}\right|_{r=R}=0. ]

The permissible eigenvalues are then determined by zeros of the derivative of the Bessel function, rather than zeros of the function itself.

The same principle applies in rectangular geometries, where zero-flux conditions select cosine-type modes or derivative-based characteristic equations.

Robin conditions create transcendental characteristic equations

A convective wall condition has the form

[ -k\left.\frac{d\Phi}{dr}\right|_{r=R}

h\Phi(R), ]

or an equivalent dimensionless form. In a cylinder, substituting the Bessel radial solution produces a characteristic equation involving both a Bessel function and its derivative.

These eigenvalues generally must be computed numerically. That is expected and does not invalidate the analytical expansion.

Preserve Orthogonality Correctly

Use the appropriate inner product

Rectangular eigenfunctions are typically orthogonal under the standard Cartesian inner product,

[ \langle f,g\rangle

\int f(x)g(x),dx. ]

For multidimensional rectangular domains, the inner product extends over the full volume.

Cylindrical radial functions are orthogonal with the radial weight (r):

[ \langle f,g\rangle_r

\int_0^R r,f(r)g(r),dr. ]

This weight is essential because it comes from the cylindrical volume element (r,dr,d\theta,dz).

Normalize modes before calculating coefficients

After identifying the eigenfunctions, normalize them so that

[ \langle \Phi_i,\Phi_j\rangle=\delta_{ij}, ]

where (\delta_{ij}) is the Kronecker delta.

The expansion coefficient for an initial temperature, concentration, or forcing distribution (u) is then obtained from a scalar projection,

[ c_i=\langle u,\Phi_i\rangle. ]

For cylindrical models, this projection must include the correct radial weight and, where applicable, integration over angle and axial position.

Include the zero mode when the physics permits it

Pure Neumann problems can contain a constant eigenfunction with (\lambda=0). This mode represents conservation of the spatial average when there is no net heat or mass flux through the boundaries.

Omitting it can produce an incorrect long-time solution, especially for insulated solids or closed diffusion systems.

Understanding the Trade-offs

Rectangular bases are simpler to implement

Sine and cosine bases have simple eigenvalues and normalization formulas for standard rectangular boundary conditions. They are often the most efficient choice for flat plates, slabs, rectangular ducts, and Cartesian computational domains.

Their simplicity decreases when different boundaries use different Robin parameters or when material properties vary spatially.

Cylindrical bases are physically accurate but more involved

Bessel functions correctly represent radial spreading and the changing cross-sectional area of cylindrical equipment. Replacing them with Cartesian sine functions can distort the radial solution, particularly near the axis or when radial gradients dominate.

The additional complexity is the need to solve Bessel characteristic equations, apply radial weighting, and handle regularity at (r=0).

Nonmatching functions create boundary and coefficient errors

A basis that does not satisfy the required homogeneous boundary conditions leaves unwanted boundary terms after integration by parts. The transformed diffusion or conduction equation then contains unresolved terms or incorrect forcing contributions.

This is not merely a cosmetic loss of elegance. It can change the predicted surface flux, transient decay rates, and long-time equilibrium.

Ideal eigenfunction methods have modeling limits

The standard sine, cosine, and Bessel constructions assume a linear operator with suitable properties, often including constant or sufficiently regular material properties. Strongly variable properties, irregular geometry, nonlinear transfer coefficients, or complicated internal sources may require numerical methods such as finite volumes, finite elements, or numerical spectral techniques.

The geometry-based basis selection remains useful in those cases, but it may serve as a benchmark, reduced model, or locally appropriate approximation rather than a complete exact solution.

How to Apply This to Your Pilot Plant

Begin by homogenizing the boundary conditions and writing the spatial operator in the coordinate system that matches the equipment.

If your primary focus is a rectangular slab, channel, or flat-plate unit: Use sine and cosine eigenfunctions, taking products across coordinates and selecting each factor to satisfy the corresponding Dirichlet, Neumann, or Robin boundaries.
If your primary focus is a solid cylindrical or tubular unit: Use Bessel functions for radial modes, discard singular radial solutions at the axis, and combine them with axial sine or cosine modes.
If your primary focus is an annular or shell-side region: Allow both regular and singular Bessel-function branches where the domain excludes the axis, then enforce both radial wall conditions to determine their combination and eigenvalues.
If your primary focus is an axisymmetric model: Retain only the zero angular mode and use the radial weight (r) when enforcing orthogonality and calculating coefficients.
If your primary focus is convective wall transfer: Form the characteristic equation from the Robin condition and compute its eigenvalues numerically rather than forcing a standard Fourier sequence.
If your primary focus is conservation in insulated equipment: Check explicitly for the zero eigenvalue and include the constant mode when the boundary conditions permit it.

Choosing a basis that matches both the pilot plant geometry and its boundary conditions turns the transport model into a valid and efficient modal representation.

Summary Table:

Pilot plant geometry	Recommended basis functions	Key considerations
Rectangular slab, channel, or flat plate	Sine and cosine functions, combined as coordinate products	Select each factor according to Dirichlet, Neumann, or Robin boundaries.
Solid cylinder or tubular unit	Bessel functions radially with axial sine or cosine functions	Exclude singular Bessel solutions at the axis and apply the correct radial eigenvalues.
Annular or shell-side region	Combinations of Bessel functions of the first and second kind	The domain excludes the axis, so both radial branches may be required.
Axisymmetric equipment	Zero angular Fourier mode	Use the radial weight `r` for orthogonality and coefficient calculations.
Convective walls	Geometry-specific eigenfunctions with numerically calculated Robin eigenvalues	Apply the flux-to-value relation to form the characteristic equation.
Insulated equipment	Cosine or derivative-based modes, including the constant mode when applicable	Retain the zero eigenvalue to preserve spatial-average conservation.

Planning a pilot-scale transport or separation system? Contact LABPARK for expert support in selecting and applying the right Educational and Vocational Unit Operations Pilot Plants. LABPARK supplies chemical engineering, bioprocess and biotech, and environmental and water treatment pilot plants for universities, research institutes, and enterprises. Our solutions help you model real equipment accurately, validate diffusion and conduction behavior, train operators and students, and accelerate process research with geometry-appropriate designs, dependable instrumentation, and practical technical guidance.