Novel Method of Fast Measurement of Thermal Conductivity in Wide Range of Temperatures

Abstract

Theoretical aspects and practical realization of a new method of fast measurement of thermal conductivity of coating materials (like deposits on wall and tubes of the boilers using fossil fuel, or high-temperature ceramics and adhesive materials) are presented. The method is based on analytical solution of the problem of unsteady heat transfer from the heated probe with coating to the environmental air. Several models of the phenomenon, taking into account convection and thermal radiation components of heat transfer, are described. The records of the probe temperature are processed numerically and compared with calculated values based on analytical solution. The best fit between the curves of recorded and calculated temperatures yields thermal conductivity value used as a parameter of the best fitting. This approach was applied to measurement of thermal properties of Fouling deposition in a coal-firing boiler. It was also demonstrated how extended version of the method allows one to represent the thermal conductivity of coating as a piecewise constant function of temperature, using a single record of the probe cooling curve. With a specially designed software for efficient calculation according to developed mathematical model, the whole experiment in the range 1500 °K–300 °K takes no more than 35–40 min, including processing on a regular PC. Validation of the suggested method was carried out in a special experiment with the substance of known thermal conductivity (Corn oil). The method was applied to Fouling depositions of two kinds of fuel—Columbian bituminous coal and Russian bituminous coal. It was demonstrated that Fouling in these two cases is characterized by significantly different thermal conductivity which also changes markedly with temperature.

Appendices

Appendices

1.1 Appendix A

The zone II of the probe is bounded by two cylindrical surfaces, the inner of radius r_i and the outer of radius r_o. Differential equation of temperature in such a body is

$$\frac{{\partial \vartheta^{II} }}{\partial t} = k_{2 } \left[ { \frac{{\partial^{2} \vartheta^{II} }}{{\partial r^{2} }} + \frac{1}{r} \frac{{\partial \vartheta^{II} }}{\partial r} } \right]; \vartheta^{II} = T^{II} - T_{\infty }$$

with well-known partial solution

$$\vartheta^{II} \left( {r,t} \right) = \left[ {B* J_{0} \left( {\alpha r} \right) + C* Y_{0} \left( {\alpha r} \right)} \right]*{ \exp }\left( { - k_{2} \alpha^{2} t} \right)$$

(A1)

valid for some arbitrary constant values α.

Obeying the boundary condition (8) for any time t, including t = 0, we get

$${\text{B}}*\alpha *J_{1} (\alpha r_{0} ) + {\text{C}}*\alpha *Y_{1} (\alpha r_{0} ) = {\text{ H}}[{\text{B}}*J_{0} \left( {\alpha r_{0} } \right) + C* Y_{0} \left( {\alpha r_{0} } \right)]$$

which yields

$${\text{C}} = - {\text{B}}*[\alpha *J_{1} (\alpha r_{0} ) \, {-}{\text{ H}}*J_{0} \left( {\alpha r_{0} } \right)]/[\alpha *Y_{1} (\alpha r_{0} ) \, {-}{\text{ H}}*Y_{0} \left( {\alpha r_{0} } \right)].$$

(A2)

Substituting in (A1) we receive

$$\vartheta^{II} \left( {r,t} \right) = B*[J_{0} \left( {\alpha r} \right) - W(r_{0} )*Y_{0} \left( {\alpha r} \right)] *{ \exp }\left( { - k_{2} \alpha^{2} t} \right)$$

(A3)

Where

$${\text{W}}\left( {{\text{r}}_{0} } \right) = [\alpha J_{1} (\alpha r_{0} ) \, - {\text{ H }}*J_{0} \left( {\alpha r_{0} } \right)\left] / \right[\alpha Y_{1} (\alpha r_{0} ) \, - {\text{ H }}*Y_{0} \left( {\alpha r_{0} } \right)].$$

(A4)

From (A3) we get

$$\lambda_{F} \left( {\frac{{\partial \vartheta^{II} }}{\partial r}} \right)_{{r_{i} }} = \lambda_{F} * B*\alpha *\left[ * \right[J_{1} \left( {\alpha r_{i} } \right) - W(r_{0} )*Y_{1} \left( {\alpha r_{i} } \right)] *{ \exp }\left( { - k_{2} \alpha^{2} t} \right).$$

(A5)

Using the partial solution of zone I

$$\vartheta^{I} \left( {{\text{r}},{\text{t}}} \right) = {\text{A}}*J_{0} \left( {\alpha r} \right)*{ \exp }( - k_{1} \alpha^{2} t)$$

(A6)

we get the corresponding component of the heat flux as follows:

$$\lambda_{tube} \left( {\frac{{\partial \vartheta^{I} }}{\partial r}} \right)_{{r_{i} }} = A*\alpha * J_{1} \left( {\alpha r_{i} } \right)*{ \exp }( - k_{1} \alpha^{2} t).$$

(A7)

We can now use expressions (A3), (A5), (A6), and (A7) in the continuity conditions on the surface r = r_i:

$$\vartheta^{I} \left( {r_{i} ,t} \right) = \vartheta^{II} \left( {r_{i} ,t} \right);\,\lambda_{tube} \left( {\frac{{\partial \vartheta^{I} }}{\partial r}} \right)_{{r_{i} }} = \lambda_{F} \left( {\frac{{\partial \vartheta^{II} }}{\partial r}} \right)_{{r_{i} }} .$$

Dividing the left side by left side and right side by right side of the above conditions, we get the equation with regard to parameter α $r_{i} = \mu$:

$$\beta *{\text{D}}_{ 1} *{\text{J}}_{0} (\mu ) \, - {\text{D}}_{ 2} *{\text{J}}_{ 1} (\mu ) \, = \, 0,$$

(A8)

where ${\text{D}}_{ 1} = \, \mu [J_{1} \left(\upmu \right)*Y_{1} \left( {m\upmu} \right) - J_{1} \left( {{\text{m}}\upmu} \right)*Y_{0} \left(\upmu \right)\left] { + H} \right[J_{0} \left( {{\text{m}}\upmu} \right)*Y_{1} \left( {m\upmu} \right) - J_{1} \left(\upmu \right)*Y_{0} \left( {{\text{m}}\upmu} \right)]$; ${\text{D}}_{ 2} = \,\upmu[J_{0} \left(\upmu \right)*Y_{1} \left( {{\text{m}}\upmu} \right) - J_{1} \left( {{\text{m}}\upmu} \right)*Y_{0} \left(\upmu \right)\left] { + H} \right[J_{0} \left( {{\text{m}}\upmu} \right)*Y_{0} \left(\upmu \right) - J_{0} \left(\upmu \right)*Y_{0} \left( {{\text{m}}\upmu} \right)]$; β = $\lambda_{F} /\lambda_{tube}$; and m = $r_{0} /r_{i}$

All roots of the equation (A8) are the eigenvalues of the differential equations for zone I and II, so all of them should be included in the summation according to expressions (5), (7), and (9) given in Sect. 3.1.

1.2 Appendix B

For the thin rod with heat sources f(x, t) = $\frac{{2\pi r_{i} }}{{Cp_{tube} * \pi r_{i}^{2} }}H*(T^{I} - T_{\infty }$) (see [5], chapter IV), we have the following equation related to the temperature $\vartheta \left( {x,t} \right) = T\left( {x,t} \right) - T_{\infty }$:

$$\frac{\partial \vartheta }{\partial t} = k_{1 } \left[ { \frac{{\partial^{2} \vartheta }}{{\partial x^{2} }} } \right] - f\left( {x,t} \right)$$

(B1)

which by substitution

$\vartheta = u*{ \exp }\left( { - \gamma t} \right)$:

is transferred into equation

$$\frac{\partial u}{\partial t} = k_{1 } \left[ { \frac{{\partial^{2} u}}{{\partial x^{2} }} } \right]$$

(B2)

with initial condition

$${\text{u}}\left( {{\text{x}},0} \right) = {\text{ T}}_{\text{B}} - {\text{T}}_{\infty }$$

(B3)

and boundary conditions

$${\text{X}} = 0:\frac{\partial u}{\partial x} = \frac{{h_{1}^{'} }}{{\lambda_{tube} }}u\left( {0,{\text{t}}} \right)$$

(B4a)

$${\text{X}} = {\text{L}}: - \frac{\partial u}{\partial x} = \frac{{h_{2}^{'} }}{{\lambda_{tube} }}u\left( {L,t} \right)$$

(B4b)

(here we took into account that convection and radiation on both surfaces of the rod can differ from each other).

Again presenting u as a product of two functions u(x,t) = X(x) T(t), we separate the variables using well-known technique of Fourier series which renders the following:

$$u\left( {x,t} \right) = \mathop \sum \limits_{n} A_{2n} \left[ {cos\mu_{n} \frac{x}{L} + \frac{{h_{1}^{'} }}{{\lambda_{tube} }}\frac{L}{{\mu_{n} }}sin\mu_{n} \frac{x}{L}} \right]\exp \left[ { - \left( {\eta \mu_{2n}^{2} ]} \right)t} \right],$$

(B5)

Where

$$A_{2n} = I1/I2;$$

(B6)

$${\mathbf{I}} 1= \mathop \smallint \limits_{0}^{L} \left[ {{ \cos }\left( {\mu_{2n} \frac{x}{L}} \right) + \frac{H1}{{\mu_{2n} }}sin(\mu_{2n} \frac{x}{L})} \right]{\text{dx}} = \frac{ L}{{\mu_{2n} }} *\left[ {sin\mu_{2n} + \frac{H1}{{\mu_{2n} }}\left( {1 - cos\mu_{2n} } \right)} \right];$$

$${\text{I2}} = \mathop \smallint \limits_{0}^{L} \left[ {{ \cos }\left( {{{\mu }}_{2n} \frac{x}{L}} \right) + \frac{H1}{{{{\mu }}_{2n} }}\sin \left( {{{\mu }}_{2n} \frac{x}{L}} \right)} \right]^{2} {\text{dx}} = \frac{L}{2}\left( {1 + \left( {\frac{H1}{{{{\mu }}_{2n} }}} \right)^{2} } \right) + \left( {1 - \left( {\frac{H1}{{{{\mu }}_{2n} }}} \right)^{2} } \right) * \left( {\sin {{\mu }}_{2n} } \right)/4{{\mu }}_{2n} + \frac{H1}{{2{{\mu }}_{2n}^{2} }}(1 - \cos {{\mu }}_{2n} )$$

with ${{\mu }}_{2n }$ to be positive roots of equation

$${\text{ctg}}{\mu}_{2n} = \frac{{{{\mu }}_{2n} }}{h1 + h2} - \frac{h1 *h2}{{{{\mu }}_{2n} \left( {h1 + h2} \right)}}$$

(B7)

Therefore, solution of ϑ from (B1) (and finally expression for T(x,t)) is as follows:

$${\text{T}}\left( {{\text{x}},{\text{t}}} \right) = {\text{ T}}_{\infty } + \left( {{\text{T}}_{\text{B}} - {\text{T}}_{\infty } } \right)\mathop \sum \limits_{n} A_{2n} \left[ {cos{{\mu }}_{2n} \frac{x}{L} + \frac{{h_{1}^{ '} }}{{\lambda_{tube} }} *\frac{L}{{{{\mu }}_{2n} }}sin{{\mu }}_{2n} \frac{x}{L}} \right] \exp \left[ { - \left( {\eta \mu_{2n}^{2} + \gamma } \right)t} \right]$$

(B8)

with the elements defined in (B6)–(B7) and (10), (11) in Sect. 3 and we also keep in mind that solutions of equation (B2) corresponding to different roots of (B7) obey the orthogonality conditions.

1.3 Appendix C

Let the full record of the probe covers the temperature range from T_B to T∞ and let us divide it into q intervals (segments): the first being from T_B to T₁ with time duration ∆t₁ indexed as q = 1, the second being from T₁ to T₂ with time duration ∆t₂ indexed as q = 2, etc. In each segment equations (C1) and (C2) are remained valid as well as both boundary conditions; however, the initial condition (T(r,0) = T_B) is valid only for q = 1, whereas for the second segment the initial condition is

$$T_{2}^{I} \left( {r,0} \right) = {\text{ T}}_{\infty } + \left( {{\text{T}}_{\text{B}} - {\text{T}}_{\infty } } \right)\sum\limits_{n} {B_{n} [\exp \left[ { - \left( {\frac{{k_{av} }}{{r_{0}^{2} }}\mu_{2n}^{2} ]} \right)\Delta t_{1} } \right]}$$

(C1)

where B_n = $\{ 2A*J_{0} (\mu_{n}$ /m)/[($\mu_{n}^{2} + A^{2} )*J_{0} \left( \mu \right]$ }* and ${{\mu }}_{2n}$ are the roots of (17) (or (B7) or (A8)). Following the general theory we represent solution of equation in segment q = 2 as a Fourier series similar to what was used in the 1^st segment, but the eigenvalues and eigenfunctions are different from what was used for segment 1 (because parameters of convection (h₂) and radiation (T_i and emissivity) were changed at transition from step 1 to the step 2). For example, we get

$$T_{2}^{I} \left( {r ,t} \right) = {\text{T}}_{\infty } + \mathop \sum \limits_{m} B_{m} \left[ { r\left[ {J_{0} \left( {\mu_{m} r} \right)} \right]} \right][\exp \left[ { - \left( {\eta \mu_{m}^{2} ]} \right)\left( {\Delta t_{1} + t} \right)} \right] ,$$

(C2)

where $\mu_{m}$ are the roots of Eq. (17) (or equivalent, if another model is exploited) with new parameters in boundary conditions and coefficients B_m are defined by expansion of initial conditions (C1) into Fourier series of new eigenfunctions (which all are orthogonal to each other since they are partial solutions of Eqs. (1) and (2)). Following the standard procedure we have

$$B_{m} = \frac{{\mathop \smallint \nolimits_{0}^{{r_{0} }} \vartheta_{2}^{I} \left( {r,0} \right)r\left[ {J_{0} \left( {\mu_{m} r} \right)} \right]*dr}}{{\mathop \smallint \nolimits_{0}^{{r_{0} }} r\left[ {J_{0} \left( {\mu_{m} r} \right)} \right]^{2} dr}} = Int1_{q} \left( {\mu_{m} } \right)/Int2_{q} \left( {\mu_{m} } \right).$$

(C3)

To calculate the first integral in (C3) we divide the r₀ on p intervals and find for each coordinate r_j = r₀*j/p, (j = 0, 1,…p−1) the temperature at the last moment of the first segment, $T_{1}^{I} \left( {r_{j} ,\Delta t_{1} } \right) - T_{\infty } = \vartheta_{2}^{I} \left( {r_{j} , 0} \right)$ as per (C1). Then, using standard way of integral calculation we get

$$Int1_{q} (\mu_{m} ) = \left( {\delta r} \right)*\mathop \sum \limits_{j = 0}^{j = p - 1} \vartheta_{2}^{I} (r_{j} ,0)r_{j} \left[ {J_{0} \left( {\mu_{m} r_{j} } \right)} \right] ,$$

(C4)

where $\delta r = r_{0} /p$. And for the second integral in (C3) we have

$$Int2_{q} (\mu_{m} ) = \frac{{r_{0}^{2} }}{{\mu_{m}^{2} }} \left\{ {\left( {A^{2} + \mu_{m}^{2} } \right)[J_{0 } \left( {\mu_{m} } \right]^{2} } \right\}$$

(C5)

with A = $r_{0} *h_{2}^{' } .$

Substitution of (C4) and (C5) in (C3) and then B_m in (C2) allows one to find temperature in segment 2 at each location r and time t.

Proceeding further in a similar way one can find the temperature distribution in segment q = 3, and then for q = 4, etc. It should be kept in mind that time-dependent component of segment q (the exponents of (C2) and corresponding expression for the other segments) comprising the sum ($\Delta t_{1} + \Delta t_{2} + \ldots + \Delta t_{q - 1} )t$ is related to all previous segments.