© 2026 The authors. This article is published by IIETA and is licensed under the CC BY 4.0 license (http://creativecommons.org/licenses/by/4.0/).
OPEN ACCESS
The particle material of a water-based nanofluid sets the balance between the gain in convective heat transfer and the hydraulic penalty, but the two effects are seldom reported on a common basis. This study compares CuO, Al₂O₃ and TiO₂ water-based nanofluids in a double-pipe heat exchanger under identical geometric and flow conditions. Effective properties were obtained from mixture rules, the Brinkman viscosity relation and a Brownian-corrected Koo–Kleinstreuer conductivity model. Heat transfer was evaluated with the Gnielinski correlation and the effectiveness–NTU method. Irreversibility was quantified with an entropy-generation balance. The MATLAB model was verified against an independent Python implementation and compared with published data. Volume fractions of 0–5%, particle diameters of 30–60 nm and Reynolds numbers of 5,000–50,000 were examined. At a volume fraction of 5% and Re = 20,000, Al₂O₃ gave the largest convective-coefficient enhancement (+13.31%) and the largest pressure-drop increase (+12.37%). CuO gave the smallest pressure-drop increase (+1.88%) and the highest effectiveness (38.65%), and it was the only nanofluid that reduced total entropy generation. CuO also ranked first on the coefficient-based performance criterion (1.1316), followed by Al₂O₃ (1.0941) and TiO₂ (1.0824). For the geometry and conditions examined, the hydraulic penalty therefore discriminates between the materials more strongly than the thermal gain.
double-pipe heat exchanger, entropy generation, Koo–Kleinstreuer model, MATLAB modelling, nanofluid, thermal–hydraulic performance
Heat exchangers are central to energy conversion and recovery in the process, power and food industries, and improving their performance remains one of the most direct routes to lower energy consumption [1, 2]. Two complementary strategies have been pursued: modifying the geometry of the flow passage and replacing the conventional working fluid with a nanofluid. A nanofluid is a colloidal suspension of solid particles of 1–100 nm in a base liquid such as water [2, 3].
Dispersed metal, metal-oxide or carbon-based particles alter the thermophysical properties of the base fluid. The thermal conductivity changes most strongly, but the specific heat and viscosity change as well, so any gain in heat transfer is accompanied by a change in hydraulic behaviour [3]. These property changes have motivated the use of nanofluids in heat exchangers for energy-conversion systems [4], and recent property measurements have extended the range of candidate fluids to multi-component (hybrid) suspensions [5]. Studies [4, 5] presented general background only; neither addressed a straight double-pipe exchanger or a single-material comparison of the type considered in this study.
Most studies of nanofluid exchangers combine a fluid-side modification with a geometric one. Ghazanfari et al. [6] showed by computational fluid dynamics (CFD) that both twisted tubes and Al₂O₃, Cu, CuO or TiO₂ nanofluids (0.05–0.15%) improved the thermal performance of a shell-and-tube exchanger. In double-pipe exchangers, Jasim et al. [7] combined a steel-ball porous medium with an Al₂O₃/water nanofluid. The overall heat-transfer coefficient, the number of transfer units (NTU) and the effectiveness all increased with flow rate and concentration. Jassim and Ahmed [8] measured Nusselt-number gains of up to 23% for Cu/water and 13% for Al₂O₃/water at 0.26–0.83% and Re = 7,000–15,000. They also observed a heat leakage that grew with concentration. Kannadhasan et al. [9] observed a marked increase in heat-transfer rate when a 0.3 wt.% CuO/water nanofluid was used as the cold stream of a counter-flow exchanger.
Fins, inserts and alternative geometries have also been studied. A curved rectangular fin raised the convective coefficient of Al₂O₃ and TiO₂ nanofluids by up to 85% with a lower pressure drop than a straight fin [10]. A helical-coil double-pipe exchanger gave coefficient and Nusselt-number ratios of 1.43 and 1.38 for 1.0 vol.% Al₂O₃/water [11]. Twisted tapes in a porous duct with a hybrid MWCNT–Al₂O₃/water nanofluid improved the Nusselt number by up to 35.6% [12]. In addition, a triangular tube under a non-uniform magnetic field raised the Nusselt number of an Fe₃O₄/water ferrofluid by about 15% [13].
Direct comparisons of particle materials are less common. Salim et al. [14] compared TiO₂, Al₂O₃ and Fe₃O₄ water-based nanofluids (1–10%) in an N-shape double-pipe exchanger. Al₂O₃ gave the highest effectiveness, and the Nusselt number increased by more than 55% at 6%. Ajeeb et al. [15] tested Al₂O₃ nanofluids (0.01–0.20%) in a compact plate exchanger and reported a 27.8% gain in heat-transfer rate at the cost of an 8% increase in pressure drop.
Recent work has refined both the modelling and the material set. Alhulaifi [16] reported a 3.5% gain in the convective coefficient of TiO₂/water (0.2–0.6 vol.%) in a counter-flow double-pipe exchanger at Re = 4,000–18,000. Other studies examined Fe₃O₄/water against condensing R-134a vapour (up to 13.4%) [17] and CuO nanofluids with baffles and turbulators (Nusselt-number gains of up to about 350%) [18]. Gains of 35% and of up to 48.8% were reported for an MgO–carboxymethyl-cellulose suspension in a U-bend exchanger [19] and for Al₂O₃ and CuO nanofluids in annular curved tubes [20], respectively. An MnCoFeO₄/water nanofluid optimised by response-surface methodology gave a 40% gain with a 15.4% rise in friction [21], and Al₂O₃, CuO and hexagonal boron-nitride nanofluids at 5–10% were analysed with CFD and an artificial neural network [22].
The studies reviewed above differ in their relevance to the present configuration. Studies [6, 7, 10-13, 17-21] involved at least one feature absent from the present configuration: another exchanger type, a curved, helical or U-bend passage, inserts, porous media, a magnetic field, cold-side phase change or a hybrid suspension. They illustrate the range of available enhancement mechanisms, but their results are not directly comparable with those of a smooth, straight double-pipe exchanger. Studies [8, 9, 16] examined straight double-pipe exchangers with single-material water-based nanofluids and are therefore the closest counterparts of the present work. Studies [14, 15, 22], in contrast, compared several particle materials in other geometries or at other concentrations.
Three gaps follow from this review. First, most studies cover a narrow concentration range, so the sensitivity of the response to volume fraction is not resolved systematically. Second, the thermal gain and the hydraulic penalty are usually reported separately, although it is their balance that determines whether a nanofluid is beneficial. Combined thermal–hydraulic and second-law assessments on a common basis remain rare. Third, the predicted enhancement depends strongly on the conductivity model, yet the model is seldom stated, which hinders comparison between studies.
The present study addresses these gaps for a smooth, straight double-pipe exchanger. CuO, Al₂O₃ and TiO₂ water-based nanofluids are compared in a single configuration over wide ranges of volume fraction, particle diameter and Reynolds number, using the Brownian-corrected Koo–Kleinstreuer conductivity model. The convective-coefficient enhancement, the pressure-drop penalty, two performance criteria and the entropy generation are reported on a common basis. The study makes two contributions. First, a like-for-like ranking on thermal, hydraulic, energy-efficiency and second-law criteria shows that, for the geometry and operating conditions considered, the material with the largest thermal enhancement is not the best overall choice. Second, a compact closed-form correlation for the convective-coefficient ratio is proposed for design estimates within its stated validity range.
2.1 Geometry and operating conditions
The exchanger consists of two concentric copper tubes (Figure 1). The nanofluid flows in the inner tube as the hot stream, and cooling water flows in the annulus as the cold stream. Both counter-flow and parallel-flow arrangements were considered. The geometric parameters are listed in Table 1. The dimensions lie at the lower end of the standard range for double-pipe hairpins [23]. In this range, the outer pipe has an inner diameter of 50–400 mm, the inner tube an outer diameter of 19–100 mm and each section a nominal length of 1.5–12 m. The resulting length-to-diameter ratio of 182 is close to the ratios of 131–200 used in the experiments of Aghayari et al. [24]. The operating conditions and the parameter ranges are summarised in Table 2.
The hot-side inlet temperature (50–90 ℃) is distinct from the temperature at which the properties were evaluated. The evaluation temperature is the bulk-mean value Tavg = (Th,in + Tc,in)/2. It equals 45 ℃ (318.15 K) in the reference case and ranges from 35 to 55 ℃ (308.15–328.15 K) over the inlet-temperature sweep. This range lies inside the 298–363 K validity interval of the adopted conductivity model [25, 26]. The hot-side inlet temperatures themselves (323.15–363.15 K) also fall inside this interval. The non-zero volume fractions (1–5%) lie within the 1–10% interval over which the Brownian coefficients were fitted by Vajjha and Das [26].
|
Parameter |
Symbol |
Value |
|
Inner tube, inner diameter |
Di,in |
22 mm |
|
Inner tube, outer diameter |
Di,out |
25 mm |
|
Outer tube, inner diameter |
Do,in |
50 mm |
|
Outer tube, outer diameter |
Do,out |
55 mm |
|
Exchanger length |
L |
4.0 m |
|
Annulus hydraulic diameter |
Dh |
25 mm |
|
Inner heat-transfer area |
Ai |
0.2765 m² |
|
Outer heat-transfer area |
Ao |
0.3142 m² |
|
Length-to-diameter ratio |
L/Di,in |
182 |
|
Wall material / conductivity |
kwall |
Copper, 385 W m⁻¹ K⁻¹ |
Table 2. Operating conditions and study parameters
|
Parameter |
Symbol |
Value / Range |
|
Hot-side inlet temperature |
Th,in |
70 ℃ (50–90 ℃ studied) |
|
Cold-side inlet temperature |
Tc,in |
20 ℃ |
|
Bulk-mean (evaluation) temperature |
Tavg |
45 ℃ (35–55 ℃) |
|
Cold-side mass flow rate |
ṁc |
0.25 kg s⁻¹ |
|
Reynolds number (inner tube) |
Re |
5,000; 10,000; 20,000; 35,000; 50,000 |
|
Volume fraction |
φ |
0, 1, 2, 3, 4, 5% |
|
Particle diameter |
dp |
30, 45, 60 nm |
|
Particle material |
– |
Al₂O₃, CuO, TiO₂ |
|
Flow arrangement |
– |
Parallel, counter |
|
Total number of cases |
– |
460 (450 nanofluid + 10 base fluid) |
Note: Tavg = (Th,in + Tc,in)/2; the value in brackets is its range over the inlet-temperature sweep. The case count is derived in Section 4.
The base-fluid properties in Table 3 correspond to water at 40 ℃. They were held constant at these values in all calculations and were not interpolated to Tavg. The evaluation temperature therefore enters the model only through the Brownian term of Eq. (6), that is, through the factor (kBT)1/2 and the function g(T, φ) of Eq. (8). This convention was applied uniformly to all tables, figures and captions. A check with water properties at 45 ℃ changed the relative enhancements of the convective coefficient and of the pressure drop at φ = 5% by less than 0.3 percentage points and did not change the ranking of the three materials.
Table 3. Properties of the base fluid and the nanoparticles
|
Material |
ρ (kg m⁻³) |
cp (J kg⁻¹ K⁻¹) |
k (W m⁻¹ K⁻¹) |
dp (nm) |
|
Water (40 ℃) |
992.0 |
4178 |
0.631 |
– |
|
Al₂O₃ |
3970 |
765 |
40.0 |
30 / 45 / 60 |
|
CuO |
6320 |
540 |
76.5 |
30 / 45 / 60 |
|
TiO₂ |
4250 |
686 |
8.4 |
30 / 45 / 60 |
Note: The viscosity of water at 40 ℃, 0.653 mPa s, was used in Eq. (3). The water properties were held at these values in all calculations (Section 2.1).
2.2 Assumptions
The analysis was based on the assumptions that are standard for single-phase treatments of nanofluid exchangers [14, 27, 28]. The flow in both streams was taken as steady, incompressible, turbulent and fully developed. The suspension was treated as a uniformly dispersed, single-phase Newtonian fluid with no slip between the particles and the base fluid. The properties were held constant, with the base-fluid properties taken from Table 3 [25, 26, 29] and the Brownian term evaluated at Tavg. Axial conduction and heat loss to the surroundings were neglected. The inner-tube flow was specified by its Reynolds number, and the annulus flow by its mass flow rate.
3.1 Nanofluid properties
The density and the specific heat follow the mixture rule of Pak and Cho [29] and the thermal-equilibrium formulation of Xuan and Roetzel [30], respectively. In the latter, the constituents are weighted by mass rather than by volume:
${{\rho }_{\text{nf}}}=\left( 1-\varphi \right){{\rho }_{\text{bf}}}+\varphi {{\rho }_{\text{p}}}$ (1)
${{c}_{\text{p},\text{nf}}}=\frac{\left( 1-\varphi \right){{\left( \rho {{c}_{\text{p}}} \right)}_{\text{bf}}}+\varphi {{\left( \rho {{c}_{\text{p}}} \right)}_{\text{p}}}}{{{\rho }_{\text{nf}}}}$ (2)
The viscosity is obtained from the Brinkman relation [31], which remains the most widely used expression for dilute suspensions:
${{\mu }_{\text{nf}}}=\frac{{{\mu }_{\text{bf}}}}{{{\left( 1-\varphi \right)}^{2.5}}}$ (3)
Eq. (3) depends only on the volume fraction. All three nanofluids therefore have the same viscosity at a given φ. This equality is a consequence of the adopted model and not an independent material result.
The effective thermal conductivity requires closer attention, because the choice of model has a first-order effect on the predicted enhancement. The classical model of Maxwell [32], later extended to non-spherical particles by Hamilton and Crosser [33], depends only on the volume fraction and the conductivity ratio. It contains neither a length scale nor the temperature and therefore cannot represent the Brownian contribution. Koo and Kleinstreuer [25] resolved this limitation by splitting the conductivity into a static part and a Brownian part:
${{k}_{\text{nf}}}={{k}_{\text{static}}}+{{k}_{\text{Brownian}}}$ (4)
${{k}_{\text{static}}}={{k}_{\text{bf}}}\,\frac{{{k}_{\text{p}}}+2{{k}_{\text{bf}}}-2\varphi \left( {{k}_{\text{bf}}}-{{k}_{\text{p}}} \right)}{{{k}_{\text{p}}}+2{{k}_{\text{bf}}}+\varphi \left( {{k}_{\text{bf}}}-{{k}_{\text{p}}} \right)}$ (5)
${{k}_{\text{Brownian}}}=5\times {{10}^{4}}\,\beta \,\varphi \,{{\rho }_{\text{bf}}}\,{{c}_{\text{p},\text{bf}}}\sqrt{\frac{{{k}_{\text{B}}}T}{{{\rho }_{\text{p}}}{{d}_{\text{p}}}}}\ g\left( T,\varphi \right)$ (6)
Eq. (5) is the Maxwell result, and Eq. (6) is the Brownian term, with T in K and dp in m. The coefficient β and the function g(T, φ) were taken from the correlations of Vajjha and Das [26]. The function is written here as g to avoid confusion with the friction factor f. The implemented expressions, with φ as a fraction, are:
$\begin{array}{r}\beta_{\mathrm{Al}_2 \mathrm{O}_3}=8.4407(100 \varphi)^{-1.07304} \\ \beta_{\mathrm{CuO}}=9.8810(100 \varphi)^{-0.9446}\end{array}$ (7)
$\begin{aligned} g(T, \varphi) & =\left(2.8217 \times 10^{-2} \varphi+3.917 \times 10^{-3}\right) \frac{T}{T_0} -\left(3.0669 \times 10^{-2} \varphi+3.91123 \times 10^{-3}\right)\end{aligned}$ (8)
where T0 = 273 K. Vajjha and Das [26] did not report a coefficient for TiO₂–water. The Al₂O₃ expression for β was therefore used for TiO₂ as an approximation. This substitution affects only the Brownian part of the TiO₂ conductivity. Consequently, all TiO₂ results that depend on knf depend on the substitution, whereas the TiO₂ density, specific heat and viscosity do not. At φ = 3%, dp = 30 nm and 45 ℃, the Brownian part accounts for 0.0682 W m⁻¹ K⁻¹ of the TiO₂ conductivity of 0.7460 W m⁻¹ K⁻¹ (9.1%). Because dp appears in the denominator under the square root of Eq. (6), the model predicts a stronger enhancement for finer particles, a trend also reported numerically by Saha and Paul [28].
3.2 Heat transfer and pressure drop
The Reynolds and Prandtl numbers of the inner-tube flow are formed from the nanofluid properties:
$Re=\frac{4{{{\dot{m}}}_{\text{h}}}}{\pi {{D}_{\text{i},\text{in}}}{{\mu }_{\text{nf}}}},Pr=\frac{{{\mu }_{\text{nf}}}{{c}_{\text{p},\text{nf}}}}{{{k}_{\text{nf}}}}$ (9)
The Nusselt number was obtained from the Gnielinski correlation [34] rather than the older Dittus–Boelter form [35]. The Gnielinski correlation covers the whole Reynolds-number range considered here and agrees well with single-phase simulations of this geometry [27]. The Darcy friction factor was obtained from the Petukhov relation [36]:
$Nu=\frac{\left( f/8 \right)\left( Re-1000 \right)\,Pr}{1+12.7\sqrt{f/8}\,\left( P{{r}^{2/3}}-1 \right)}$ (10)
$f={{\left( 0.790\text{ln}Re-1.64 \right)}^{-2}}$ (11)
The pressure drop and the pumping power of the inner-tube stream follow from the Darcy–Weisbach relation. Because the comparison was made at a fixed Reynolds number, the mass flow rate and the mean velocity follow from Eq. (9) as ṁh = πDi,inμnfRe/4 and v = Re μnf/(ρnfDi,in), which gives:
$\Delta P=f\frac{L}{{{D}_{\text{i},\text{in}}}}\frac{{{\rho }_{\text{nf}}}{{v}^{2}}}{2}=\frac{f\left( Re \right)\,L\,R{{e}^{2}}\mu _{\text{nf}}^{2}}{2{{\rho }_{\text{nf}}}D_{\text{i},\text{in}}^{3}},P=\frac{{{{\dot{m}}}_{\text{h}}}\,\Delta P}{{{\rho }_{\text{nf}}}}$ (12)
At a fixed Re, the friction factor is the same for all fluids, and the pressure drop therefore scales with μnf2/ρnf. The fixed-Re basis was adopted because it preserves dynamic similarity between the fluids and is commonly used in single-phase studies of turbulent nanofluid flow in tubes (e.g., studies [28, 37]). A comparison at a fixed mass flow rate or velocity would weight viscosity and density differently and could change the hydraulic ranking.
The overall heat-transfer coefficient, based on the inner area Ai, combines the three thermal resistances in series, and the duty follows from the effectiveness–NTU method:
$\frac{1}{U{{A}_{\text{i}}}}=\frac{1}{{{h}_{\text{i}}}{{A}_{\text{i}}}}+\frac{\text{ln}\left( {{D}_{\text{i},\text{out}}}/{{D}_{\text{i},\text{in}}} \right)}{2\pi {{k}_{\text{wall}}}L}+\frac{1}{{{h}_{\text{o}}}{{A}_{\text{o}}}}$ (13)
$NTU=\frac{U{{A}_{\text{i}}}}{{{C}_{\text{min}}}},{{C}_{\text{r}}}=\frac{{{C}_{\text{min}}}}{{{C}_{\text{max}}}},Q=\varepsilon \,{{C}_{\text{min}}}\left( {{T}_{\text{h},\text{in}}}-{{T}_{\text{c},\text{in}}} \right)$ (14)
where C = ṁcp is the heat-capacity rate. The effectiveness of the counter-flow and parallel-flow arrangements is:
${{\varepsilon }_{\text{cf}}}=\frac{1-\text{exp}\left[ -NTU\left( 1-{{C}_{\text{r}}} \right) \right]}{1-{{C}_{\text{r}}}\text{exp}\left[ -NTU\left( 1-{{C}_{\text{r}}} \right) \right]}$ (15)
${{\varepsilon }_{\text{pf}}}=\frac{1-\text{exp}\left[ -NTU\left( 1+{{C}_{\text{r}}} \right) \right]}{1+{{C}_{\text{r}}}}$ (16)
The axial temperature profiles follow from the local energy balances of the two streams with the constant U of Eq. (13):
$\begin{gathered}\frac{d T_{\mathrm{h}}}{d x}=-\frac{U \pi D_{\mathrm{i}, \mathrm{in}}}{C_{\mathrm{h}}}\left(T_{\mathrm{h}}-T_{\mathrm{c}}\right) \\ \frac{d T_{\mathrm{c}}}{d x}= \pm \frac{U \pi D_{\mathrm{i}, \mathrm{in}}}{C_{\mathrm{c}}}\left(T_{\mathrm{h}}-T_{\mathrm{c}}\right)\end{gathered}$ (17)
The upper sign applies to parallel flow and the lower sign to counter flow. Because U and the properties are constant, Eq. (17) has the closed-form solution (Th − Tc)(x) = (Th − Tc)(0) exp[−UπDi,inx(1/Ch ± 1/Cc)]. The profiles therefore contain no information beyond the ε–NTU solution, and their outlet temperatures coincide with those of Eqs. (14)–(16). No property updating along the exchanger was performed.
3.3 Performance criteria and irreversibility
Two performance-evaluation criteria (PEC) were used. The first is the Nusselt-based form common in the nanofluid literature [37, 38]. The second replaces the Nusselt number by the convective coefficient and the friction factor by the pumping power of Eq. (12):
$PEC=\frac{Nu/N{{u}_{0}}}{{{\left( f/{{f}_{0}} \right)}^{1/3}}},PE{{C}_{\text{h}}}=\frac{{{h}_{\text{i}}}/{{h}_{\text{i},0}}}{{{\left( P/{{P}_{0}} \right)}^{1/3}}}$ (18)
The subscript 0 denotes the base fluid at the same Re, arrangement and inlet temperatures. The reference pumping power P0 was calculated from Eq. (12) with the water properties of Table 3. Because Re is fixed, the mass flow rate differs between each nanofluid and the base fluid (ṁh ∝ μnf), and this difference is included in P/P0. An energy-efficiency ratio was also evaluated:
$EER=\frac{Q/{{Q}_{0}}}{P/{{P}_{0}}}$ (19)
The irreversibility was quantified with an entropy balance over the whole exchanger, following Bejan [39]. The total entropy-generation rate is the sum of a thermal and a frictional contribution:
$\begin{aligned} \dot{S}_{\text {gen }} & =\dot{S}_{\text {gen, th }}+\dot{S}_{\text {gen,fr }} \\ \dot{S}_{\text {gen, th }} & =C_{\mathrm{h}} \ln \frac{T_{\mathrm{h}, \text { out }}}{T_{\mathrm{h}, \text { in }}}+C_{\mathrm{c}} \ln \frac{T_{\mathrm{c}, \text { out }}}{T_{\mathrm{c}, \mathrm{in}}}\end{aligned}$ (20)
$\dot{S}_{\text {gen,fr }}=\frac{\dot{m}_{\mathrm{h}} \Delta P_{\mathrm{h}}}{\rho_{\mathrm{nf}} \bar{T}_{\mathrm{h}}}+\frac{\dot{m}_{\mathrm{c}} \Delta P_{\mathrm{c}}}{\rho_{\mathrm{hf}} \bar{T}_{\mathrm{C}}}$ (21)
All temperatures in Eqs. (20) and (21) are absolute temperatures in K. The hot-side term of Eq. (20) is negative and the cold-side term is positive; their sum is positive. The frictional term includes both streams, with ΔPc calculated for the annulus with the same friction relation. The mean temperatures are T̄h = (Th,in + Th,out)/2 and T̄c = (Tc,in + Tc,out)/2. Replacing them by logarithmic-mean temperatures changed none of the reported entropy measures in the fourth decimal place. The Bejan number Be and the augmentation entropy-generation number Ns,a are defined as:
$Be=\frac{{{{\dot{S}}}_{\text{gen},\text{th}}}}{{{{\dot{S}}}_{\text{gen}}}},{{N}_{\text{s},\text{a}}}=\frac{{{{\dot{S}}}_{\text{gen}}}}{{{{\dot{S}}}_{\text{gen},0}}}$ (22)
A value of Ns,a below unity indicates that the nanofluid reduces the total entropy generation relative to the base fluid under the same conditions.
The model was implemented in MATLAB. Because the properties are constant and the effectiveness–NTU relations return the duty in closed form, no partial-differential solver is required, and each operating point is obtained in a single forward pass. This makes the code faster to develop and easier to audit than a general-purpose CFD model. The code comprises three parts: a property routine implementing Eqs. (1)–(8); a single-point solver returning Re, Pr, the two convective coefficients, U, NTU, ε, Q, the outlet temperatures, ΔP, P and the entropy terms of Eqs. (20)–(22); and a driver that sweeps the parameter space. No toolbox is required.
The parametric study comprised 460 operating points. The six volume fractions in Table 2 include φ = 0, at which the fluid is pure water irrespective of the particle material and diameter. The nanofluid cases therefore comprise three materials × five non-zero volume fractions × three particle diameters × five Reynolds numbers × two arrangements = 450 points. The base fluid was evaluated once for each Reynolds number and arrangement (5 × 2 = 10 points), giving 450 + 10 = 460 points in total. A further 1,575 points were generated for the correlation of Section 6.5. The complete sweep runs in a few seconds on a standard desktop computer. Two coding errors were identified and corrected during code review: the reserved keyword if had been used as a loop index, and the structure fields were ordered inconsistently between two branches of the property routine. Neither error affected the results reported here.
5.1 Independent cross-check of the MATLAB code
Before any result was interpreted, the whole procedure was transcribed independently into Python 3, and the two implementations were compared variable by variable. The test matrix comprised three materials, four volume fractions (0, 1, 3 and 5%), three particle diameters (30, 45 and 60 nm) and three Reynolds numbers (5,000, 20,000 and 50,000). It also included two flow arrangements, two conductivity models (Maxwell and Koo–Kleinstreuer) and three hot-side inlet temperatures (50, 70 and 90 ℃). This gives 3 × 4 × 3 × 3 × 2 × 2 × 3 = 1,296 operating points. At each point, 21 output quantities were compared: ρnf, cp,nf, μnf, knf, Pr, Nu, f, hi, U, NTU, ε, Q, ΔP, P, ṁh, Th,out, Tc,out, Ṡgen,th, Ṡgen,fr, Ṡgen and Be. In total, 27,216 comparisons were made. The largest relative deviation was 4.24 × 10⁻¹⁴%, at the level of double-precision round-off. This result confirms that the MATLAB code is free of transcription errors. It does not, by itself, validate the physical model.
5.2 Comparison with published data
To place the predictions in the context of published results, the model was compared with four published studies (Table 4). The comparison was made for Al₂O₃–water at φ = 3%, the only condition available in all four studies. The present value, Δh = 10.30%, corresponds to dp = 30 nm, Re = 20,000 and counter flow. The deviation is defined as |Δhpresent − Δhref|/Δhref × 100. The reference studies differ from the present one in geometry, Reynolds number, particle size or method, as listed in Table 4. The comparison is therefore presented as a consistency check and as qualitative context, not as a quantitative validation.
Table 4. Comparison of the convective-coefficient enhancement Δh for Al₂O₃–water (φ = 3%) with published studies
|
Study |
Configuration and Method |
Reference Δh (%) |
Deviation (%) |
Role in the Comparison |
|
Present (Koo–Kleinstreuer) |
Straight double-pipe, single-phase, Gnielinski; dp = 30 nm, Re = 20,000 |
+10.30 |
– |
– |
|
Present (Maxwell) |
As above, Maxwell conductivity |
+4.31 |
– |
– |
|
Zheng et al. [27] |
Straight double-pipe inner tube, single-phase, Gnielinski |
10–20 (range) |
Inside range |
Consistency check (same method) |
|
Saha and Paul [28] |
Straight circular pipe, single-phase, turbulent |
≈9.7 |
6.2 |
Consistency check (same method) |
|
Salim et al. [14] |
N-shape double-pipe (secondary flows) |
+16.77 |
38.6 |
Qualitative context |
|
Aghayari et al. [24] |
Experimental double-pipe, lower φ |
+21.5 |
52.1 |
Qualitative context |
Note: Deviation = |Δhpresent − Δhref|/Δhref × 100, with Δhpresent = 10.30%. Reference values were taken at φ = 3% where reported; the Reynolds numbers and particle sizes of the reference studies differ from those of the present case.
The study of Zheng et al. [27] is closest to the present one in method: a straight double-pipe exchanger, a single-phase treatment of the nanofluid and the same Gnielinski correlation. The present prediction of 10.30% at 3% lies inside the 10–20% range that they report, and the agreement holds at 4% and 5%. Against Saha and Paul [28], who used a circular pipe and the same single-phase approach, the deviation is 6.2%. The two remaining comparisons deviate considerably more, for structural rather than numerical reasons. Salim et al. [14] studied an N-shaped exchanger, whose bends generate secondary flows that are absent in a straight pipe, so a higher enhancement is expected. Aghayari et al. [24] reported experimental measurements, which single-phase models are known to under-predict because they do not resolve particle migration or thermophoretic drift [28]. These two studies are therefore treated as context rather than as validation (Figure 2).
Figure 2. Comparison with previous studies for Al₂O₃–water at Re = 20,000: (a) convective-coefficient enhancement reported by each study, with the deviation from the present model in brackets; (b) present model for three particle diameters against the range of Zheng et al. [27] and the values of studies [14, 28] at φ = 3%
6.1 Effective properties and the conductivity model
Table 5 lists the properties at φ = 3%. Adding particles increased the density, thermal conductivity and viscosity and decreased the specific heat. The net effect was a reduction of about 18% in the Prandtl number, because the increase in conductivity outweighed the increase in viscosity. CuO had the highest density and the lowest specific heat of the three nanofluids, whereas Al₂O₃ had the lowest density and the highest specific heat. The effective conductivity was highest for CuO (0.7635 W m⁻¹ K⁻¹) and lowest for TiO₂ (0.7460 W m⁻¹ K⁻¹), in line with the particle conductivities in Table 3. Figure 3 shows the four properties over the whole concentration range.
Figure 3. Thermophysical properties of the nanofluids: (a) density; (b) specific heat; (c) viscosity; (d) thermal conductivity (shown over φ = 1–5%, the fitted range of the Brownian coefficients)
The choice of conductivity model matters as much as the effect under study. At φ = 3%, the Maxwell expression [32] gives a conductivity enhancement of 8.8%, whereas the Koo–Kleinstreuer model [25] gives 20.0%, a factor of 2.3. The difference carries through to the convective coefficient, for which the two models predict 4.3% and 10.3%, respectively (Figure 4). An enhancement reported without its conductivity model is therefore difficult to interpret, and the model is stated explicitly throughout this work.
Table 5. Nanofluid properties at φ = 3% and dp = 30 nm
|
Fluid |
ρ (kg m⁻³) |
cp (J kg⁻¹ K⁻¹) |
μ (mPa s) |
k (W m⁻¹ K⁻¹) |
Pr |
|
Water |
992.0 |
4178.0 |
0.6530 |
0.6310 |
4.324 |
|
Al₂O₃–water |
1081.3 |
3802.1 |
0.7047 |
0.7574 |
3.537 |
|
CuO–water |
1151.8 |
3579.2 |
0.7047 |
0.7635 |
3.303 |
|
TiO₂–water |
1089.7 |
3769.4 |
0.7047 |
0.7460 |
3.560 |
Note: The Brownian term was evaluated at Tavg = 45 ℃, and the base-fluid properties were taken from Table 3. The properties were calculated from Eqs. (1)–(8); the equal viscosities follow from the Brinkman relation, Eq. (3).
Figure 4. Thermal-conductivity ratio: (a) Maxwell [32] and Koo–Kleinstreuer (K–K) [25] models for Al₂O₃–water at three particle diameters; (b) material comparison (K–K, dp = 30 nm)
6.2 Thermal–hydraulic performance
Table 6 lists the performance at Re = 20,000 in counter flow. All three nanofluids increased the convective coefficient monotonically with concentration. Al₂O₃ gave the largest convective-coefficient enhancement, from 7.72% at 1% to 13.31% at 5%, followed by TiO₂. CuO gave the smallest enhancement despite having the most conductive particles. The specific heat explains this result: CuO has the lowest specific heat, so its Prandtl number decreases the most, and this reduction partly cancels the gain in conductivity.
Table 6. Performance at Re = 20,000, dp = 30 nm, counter flow (Eqs. (9)–(16))
|
Fluid |
φ (%) |
hi |
Δh (%) |
ε (%) |
Q (W) |
ΔP (Pa) |
Δ(ΔP) (%) |
|
Water |
0 |
3499.4 |
– |
34.07 |
16,062 |
844.6 |
– |
|
Al₂O₃ |
1 |
3769.7 |
+7.72 |
35.11 |
16,433 |
862.2 |
+2.09 |
|
Al₂O₃ |
3 |
3859.8 |
+10.30 |
35.69 |
16,521 |
902.2 |
+6.83 |
|
Al₂O₃ |
5 |
3965.1 |
+13.31 |
36.19 |
16,637 |
949.0 |
+12.37 |
|
CuO |
1 |
3720.7 |
+6.32 |
35.59 |
16,301 |
842.8 |
−0.21 |
|
CuO |
3 |
3778.5 |
+7.97 |
37.24 |
16,228 |
847.0 |
+0.29 |
|
CuO |
5 |
3841.5 |
+9.77 |
38.65 |
16,184 |
860.4 |
+1.88 |
|
TiO₂ |
1 |
3748.9 |
+7.13 |
35.13 |
16,395 |
859.8 |
+1.81 |
|
TiO₂ |
3 |
3812.4 |
+8.94 |
35.80 |
16,432 |
895.3 |
+6.01 |
|
TiO₂ |
5 |
3891.0 |
+11.19 |
36.38 |
16,500 |
937.5 |
+11.01 |
Note: hi in W m⁻² K⁻¹. The overall coefficient U is shown in Figure 5(b).
Figure 5. Variation with concentration (Re = 20,000, dp = 30 nm, counter flow; dashed line: water): (a) hi; (b) U; (c) effectiveness; (d) heat duty; (e) pressure drop; (f) entropy generation
The ordering reversed for the effectiveness: CuO reached 38.65% at 5%, compared with 36.19% for Al₂O₃. Its low specific heat reduces the minimum heat-capacity rate and thereby increases NTU. The duty followed a different ordering again. Because Q = εCmin(Th,in − Tc,in), the smaller Cmin offset the higher effectiveness, and Al₂O₃ transferred the most heat (16,637 W, compared with 16,184 W for CuO). The preferred material therefore depends on whether the design objective is maximum duty or maximum effectiveness. Figure 5 shows these quantities as functions of concentration.
6.3 Hydraulic penalty and performance criteria
The hydraulic penalty separated the three nanofluids more sharply than the thermal gain did. At φ = 5%, the pressure drop increased by 12.37% for Al₂O₃ and by 11.01% for TiO₂, but by only 1.88% for CuO. The calculation basis explains this difference. At a fixed Re, Eq. (12) gives ΔP ∝ μnf2/ρnf, because the friction factor and the geometry are common to all fluids. At a given φ, the Brinkman relation gives the same viscosity, and hence the same mass flow rate, for the three nanofluids. The differences between materials at equal φ therefore arise from the density alone, and CuO, the densest of the three, has the smallest pressure drop. This density effect is thus a consequence of the fixed-Re basis and of the Brinkman relation, not an independent hydraulic property of the material. The effect on pumping power was even larger: at 1%, CuO required 2.9% less pumping power than pure water.
The two criteria of Eq. (18) led to different conclusions. The Nusselt-based criterion was below unity for all three nanofluids, whereas the coefficient-based criterion exceeded unity for all three. This difference results from the normalisation. Because Nu = hiDi,in/knf and, at 3%, the conductivity increased by about 20% while the coefficient increased by about 10%, the Nusselt-number ratio necessarily fell even though hi increased. For a fixed exchanger geometry, the duty depends on hi through U, not on Nu. The coefficient-based criterion was therefore used to rank the nanofluids in this study. On this basis, CuO ranked first (PECh = 1.1316), followed by Al₂O₃ (1.0941) and TiO₂ (1.0824). This ranking applies to the present geometry, the fixed-Re basis and the operating conditions of Table 2. It should not be generalised to other configurations without further analysis. Figure 6 presents the three criteria.
Figure 6. Performance-evaluation criteria (Re = 20,000, dp = 30 nm, counter flow): (a) Nusselt-based, Eq. (18) [37, 38]; (b) coefficient-based, Eq. (18); (c) energy-efficiency ratio, Eq. (19)
6.4 Particle diameter and inlet temperature
Because the Koo–Kleinstreuer model retains the Brownian term, the particle diameter enters explicitly through the inverse square root in Eq. (6). Reducing the diameter of Al₂O₃ particles from 60 to 30 nm increased the conductivity enhancement from 16.72% to 20.03% and the convective-coefficient enhancement from 8.56% to 10.30% at φ = 3% and Re = 20,000. Saha and Paul [28] reported the same direction of dependence, which the classical mixture models [32, 33] cannot capture because they contain no length scale.
The Brownian term also introduces a strong temperature dependence. When the hot-side inlet temperature was raised from 50 to 90 ℃ (Tavg = 35 – 55 ℃), the conductivity enhancement of Al₂O₃–water at 5% increased from 23.4% to 29.7% and the convective-coefficient enhancement from 11.8% to 15.1%. The ordering of the three materials was preserved throughout. A classical mixture model predicts no such dependence, so this behaviour is entirely a consequence of the conductivity model.
6.5 Flow arrangement, entropy generation and correlation
The counter-flow arrangement transferred more heat in every case (Table 7): by 5.09% for pure water and by 5.57–6.02% for the nanofluids, so the margin widened slightly with concentration. The axial profiles in Figure 7, obtained from Eq. (17), explain this result: counter flow maintains a nearly uniform temperature difference along the exchanger, whereas in parallel flow the difference decreases towards the outlet.
Table 7. Effect of flow arrangement at Re = 20,000, dp = 30 nm (pf: parallel flow; cf: counter flow)
|
Fluid |
φ (%) |
εpf (%) |
εcf (%) |
Qpf (W) |
Qcf (W) |
Gain (%) |
|
Water |
0 |
32.42 |
34.07 |
15,284 |
16,062 |
+5.09 |
|
Al₂O₃ |
3 |
33.80 |
35.69 |
15,648 |
16,521 |
+5.58 |
|
Al₂O₃ |
5 |
34.23 |
36.19 |
15,736 |
16,637 |
+5.73 |
|
CuO |
3 |
35.21 |
37.24 |
15,345 |
16,228 |
+5.76 |
|
CuO |
5 |
36.46 |
38.65 |
15,265 |
16,184 |
+6.02 |
|
TiO₂ |
3 |
33.92 |
35.80 |
15,566 |
16,432 |
+5.57 |
|
TiO₂ |
5 |
34.42 |
36.38 |
15,609 |
16,500 |
+5.71 |
Table 8. Performance criteria and entropy generation at φ = 5%, Re = 20,000, dp = 30 nm, counter flow (Eqs. (18)–(22))
|
Fluid |
PECh |
EER |
Ns,a |
S* |
Be |
|
Al₂O₃ |
1.0941 |
0.9326 |
1.0117 |
0.9768 |
0.99987 |
|
CuO |
1.1316 |
1.1036 |
0.9743 |
0.9670 |
0.99989 |
|
TiO₂ |
1.0824 |
0.9477 |
1.0042 |
0.9775 |
0.99988 |
Note: Performance-evaluation criteria (PEC), energy-efficiency ratio (EER); S* = (Ṡgen/Q)/(Ṡgen,0/Q0), the entropy generation per unit duty relative to water.
Figure 7. Axial temperature distributions of water and Al₂O₃–water (φ = 5%) at Re = 20,000, obtained from Eq. (17): (a) counter flow; (b) parallel flow
With respect to the second law, Ns,a (Eq. (22)) fell below unity only for CuO, to 0.9743, a 2.6% reduction in total irreversibility relative to pure water (Table 8). Al₂O₃ and TiO₂ increased the total slightly, by 1.2% and 0.4%, respectively, because the additional duty they transferred was accompanied by additional thermal irreversibility. Per unit duty, however, all three nanofluids generated less entropy than water. For water at Re = 20,000, the thermal term was 5.39 W K⁻¹ (−48.01 W K⁻¹ on the hot side and +53.40 W K⁻¹ on the cold side), whereas the frictional term was only 6.3 × 10⁻⁴ W K⁻¹. The Bejan number therefore exceeded 0.9998 for Re ≤ 20,000 and remained above 0.9989 at Re = 50,000. The irreversibility was thus dominated by heat transfer rather than by friction over the whole range studied.
Finally, a compact correlation for the ratio of the convective coefficients was fitted separately for each material:
$\frac{{{h}_{\text{nf}}}}{{{h}_{\text{bf}}}}=1+K{{\varphi }^{a}}{{\left( \frac{{{d}_{\text{p}}}}{{{D}_{\text{i},\text{in}}}} \right)}^{b}}{{\left( \frac{{{T}_{\text{avg}}}}{{{T}_{\text{ref}}}} \right)}^{c}}$ (23)
In Eq. (23), φ is the volume fraction expressed as a fraction, dp and Di,in are both in m (Di,in = 0.022 m), Tavg is the bulk-mean evaluation temperature in K, and Tref = Tc,in = 293.15 K. The fitting data comprised 525 points per material (1,575 in total): five non-zero volume fractions (1–5%) and three particle diameters, each combined with 35 pairs of Reynolds number (5,000–50,000) and hot-side inlet temperature (50–90 ℃). The flow arrangement was not varied because hi is independent of the arrangement when the properties are evaluated at a fixed temperature. The Reynolds number was varied but omitted from Eq. (23) because its effect on the ratio is weak: at φ = 5%, hnf/hbf for Al₂O₃ changed only from 1.149 to 1.124 between Re = 5,000 and 50,000. This residual dependence is the main source of the scatter in Figure 8.
Figure 8. Proposed correlation, Eq. (23): (a) parity plot; (b) residual distribution
The coefficients are listed in Table 9. The exponent on the volume fraction is close to one third, so the enhancement grows sub-linearly with concentration. The negative exponent on the particle diameter reflects the stronger effect of finer particles, and the large positive exponent on temperature reflects the strength of the Brownian contribution. On the fitting data, the mean absolute error was below 0.7% and the maximum error below 2.3% (Figure 8). The correlation was also assessed on an independent hold-out set of 192 points per material that were not used in the fit (φ = 1.5, 2.5, 3.5 and 4.5%; dp = 35, 50 and 55 nm; Re = 7,500, 15,000, 25,000 and 42,500; Th,in = 55, 65, 75 and 85 ℃). On this set, the mean absolute errors were 0.45%, 0.53% and 0.42% and the maximum errors 1.27%, 1.35% and 1.13% for Al₂O₃, CuO and TiO₂, respectively. Both data sets were generated by the same model. The correlation therefore reproduces the model and inherits its assumptions and validity limits.
Table 9. Coefficients and accuracy of the proposed correlation (Eq. (23))
|
Fluid |
103K |
a |
b |
c |
MAE (%) |
Max (%) |
MAEho (%) |
|
Al₂O₃ |
6.515 |
0.3801 |
−0.2791 |
4.5637 |
0.566 |
2.138 |
0.447 |
|
CuO |
1.012 |
0.3026 |
−0.3683 |
6.0720 |
0.676 |
2.252 |
0.529 |
|
TiO₂ |
2.884 |
0.3195 |
−0.3100 |
5.0835 |
0.524 |
1.882 |
0.415 |
Note: Mean absolute error (MAE) and maximum (Max) refer to the fitting data and MAEho to the hold-out set. Validity: 5,000 ≤ Re ≤ 50,000; 1% ≤ φ ≤ 5%; 30 nm ≤ dp ≤ 60 nm; 50 ℃ ≤ Th,in ≤ 90 ℃ (308.15 K ≤ Tavg ≤ 328.15 K).
6.6 Overall assessment and material selection
No single material was optimal on every metric. Al₂O₃ gave the largest convective-coefficient enhancement and duty, but also the largest pressure-drop increase. TiO₂ was intermediate on all metrics. CuO gave the smallest enhancement, yet it combined the smallest hydraulic penalty with the highest effectiveness, the highest PECh and the only reduction in total entropy generation. When the thermal gain and the hydraulic penalty are weighed together, CuO is therefore the preferred working fluid for the present geometry, the fixed-Re basis and the operating conditions of Table 2.
6.7 Limitations
Four limitations should be noted. First, the single-phase treatment does not resolve particle migration or thermophoretic drift and is known to under-predict measured enhancements [24, 28]. Second, the properties were constant along the exchanger, with the base-fluid properties held at 40 ℃ (Section 2.1). Third, the Al₂O₃ Brownian coefficient was used for TiO₂ (Section 3.1). Fourth, the ranking was obtained at a fixed Reynolds number for a single geometry; other comparison bases may change it (Section 6.3).
CuO, Al₂O₃ and TiO₂ water-based nanofluids were compared in a straight double-pipe heat exchanger with a verified MATLAB model based on the Koo–Kleinstreuer conductivity model. The main conclusions are as follows:
(1) The choice of conductivity model is decisive: the Koo–Kleinstreuer model predicts more than twice the enhancement of the classical Maxwell model, so enhancement values should always be reported together with the model used.
(2) Al₂O₃ gave the largest convective-coefficient enhancement and the highest duty, whereas CuO gave the smallest enhancement but the highest effectiveness, because of its low specific heat.
(3) At a fixed Reynolds number, the pressure-drop penalty separated the nanofluids far more sharply than the thermal gain and was smallest for CuO, the densest material. This hydraulic advantage is a consequence of the comparison basis and of the Brinkman viscosity relation.
(4) CuO ranked first on the coefficient-based performance criterion and was the only nanofluid that reduced the total entropy generation relative to water. The irreversibility was dominated by heat transfer throughout.
(5) Smaller particles, higher operating temperatures and the counter-flow arrangement each increased the thermal benefit, the first two through the Brownian term, which classical mixture models cannot reproduce.
(6) For the present geometry and operating conditions, CuO offers the most favourable combined thermal–hydraulic performance. Al₂O₃ is preferable only where maximum duty is the sole objective and the pumping penalty is acceptable.
The proposed correlation, Eq. (23), can be used for rapid design estimates within its validity range. Future work should include a two-phase treatment of particle migration, temperature-dependent properties along the exchanger, a dedicated Brownian coefficient for TiO₂–water and comparisons on other bases, such as equal pumping power.
The authors thank the College of Engineering, Wasit University, and the Faculty of Engineering, University of Misan, for their academic support. This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.
|
A |
heat-transfer surface area, m² |
|
a, b, c |
exponents of Eq. (23), dimensionless |
|
Be |
Bejan number, Eq. (22), dimensionless |
|
C |
heat-capacity rate, ṁcp, W K⁻¹ |
|
Cr |
heat-capacity ratio, dimensionless |
|
cp |
specific heat capacity, J kg⁻¹ K⁻¹ |
|
D |
diameter, m |
|
Dh |
hydraulic diameter of the annulus, m |
|
dp |
nanoparticle diameter, m (reported in nm) |
|
EER |
energy-efficiency ratio, Eq. (19), dimensionless |
|
f |
Darcy friction factor, dimensionless |
|
g |
Brownian temperature function, Eq. (8), dimensionless |
|
h |
convective heat-transfer coefficient, W m⁻² K⁻¹ |
|
K |
coefficient of Eq. (23), dimensionless |
|
k |
thermal conductivity, W m⁻¹ K⁻¹ |
|
kB |
Boltzmann constant, 1.381 × 10⁻²³ J K⁻¹ |
|
L |
exchanger length, m |
|
ṁ |
mass flow rate, kg s⁻¹ |
|
Ns,a |
augmentation entropy-generation number, Eq. (22), dimensionless |
|
NTU |
number of transfer units, dimensionless |
|
Nu |
Nusselt number, dimensionless |
|
P |
pumping power, W |
|
PEC |
performance-evaluation criterion, dimensionless |
|
PECh |
coefficient-based performance-evaluation criterion, Eq. (18), dimensionless |
|
Pr |
Prandtl number, dimensionless |
|
Q |
heat-transfer rate (duty), W |
|
Re |
Reynolds number, dimensionless |
|
Ṡgen |
entropy-generation rate, W K⁻¹ |
|
S* |
entropy generation per unit duty relative to water (Table 8), dimensionless |
|
T |
temperature, K or ℃ |
|
Tavg |
bulk-mean (property-evaluation) temperature, K |
|
T0 |
reference temperature of Eq. (8), 273 K |
|
Tref |
reference temperature of Eq. (23), 293.15 K |
|
T̄ |
arithmetic-mean stream temperature, K |
|
U |
overall heat-transfer coefficient (based on Ai), W m⁻² K⁻¹ |
|
v |
mean velocity, m s⁻¹ |
|
x |
axial coordinate, m |
|
Greek symbols |
|
|
β |
Brownian-motion coefficient, Eq. (7), dimensionless |
|
ΔP |
pressure drop, Pa |
|
Δh |
relative enhancement of the convective heat-transfer coefficient, % |
|
ε |
effectiveness, dimensionless |
|
μ |
dynamic viscosity, Pa s |
|
ρ |
density, kg m⁻³ |
|
φ |
nanoparticle volume fraction, dimensionless (reported in %) |
|
Subscripts |
|
|
0 |
base fluid at the same operating conditions |
|
bf |
base fluid (water) |
|
c |
cold side |
|
cf, pf |
counter flow, parallel flow |
|
fr |
frictional |
|
h |
hot side |
|
i |
inner tube |
|
in, out |
inlet, outlet |
|
max, min |
maximum, minimum |
|
nf |
nanofluid |
|
o |
outer tube (annulus) |
|
p |
nanoparticle |
|
static, Brownian |
static (Maxwell) and Brownian contributions |
|
th |
thermal |
|
wall |
tube wall |
[1] Purbia, D., Khandelwal, A., Kumar, A., Sharma, A.K. (2019). Graphene-water nanofluid in heat exchanger: Mathematical modelling, simulation and economic evaluation. International Communications in Heat and Mass Transfer, 108: 104327. https://doi.org/10.1016/j.icheatmasstransfer.2019.104327
[2] Rashid, F.L., Al-Obaidi, M.A., Al Maimuri, N.M.L., et al. (2025). Enhancing efficiency in double-pipe heat exchangers: A detailed review of nanofluids and their impact on thermal performance. Heat Transfer, 55(1): 212-240. https://doi.org/10.1002/htj.70070
[3] Cardoso, B.D., Souza, A., Nobrega, G., et al. (2025). Progress in nanofluid technology: From conventional to green nanofluids for biomedical, heat transfer, and machining applications. Nanomaterials, 15(16): 1242. https://doi.org/10.3390/nano15161242
[4] Aleuaysh, A.Z.N. (2026). Thermal performance enhancement of heat exchangers using nano-fluid in energy conversion systems. Dijlah Journal of Engineering Sciences, 3(2): 71-95. https://doi.org/10.65204/djes.v3i2.517
[5] Nayebpashaee, N. (2026). Experimental investigation of thermophysical properties of Al2O3 TiO2 CuO Fe3O4 water ethylene glycol quadri hybrid nanofluid. Discover Nano, 21: 142. https://doi.org/10.1186/s11671-026-04602-w
[6] Ghazanfari, V., Taheri, A., Amini, Y., Mansourzade, F. (2024). Enhancing heat transfer in a heat exchanger: CFD study of twisted tube and nanofluid (Al₂O₃, Cu, CuO and TiO₂) effects. Case Studies in Thermal Engineering, 53: 103864. https://doi.org/10.1016/j.csite.2023.103864
[7] Jasim, Q.K., Saleh, N.M., Hussein, A.M. (2020). Improving thermal performance using Al₂O₃–water nanofluid in a double-pipe heat exchanger filled with a porous medium. Thermal Science, 24(6B): 4267-4275. https://thermalscience.rs/pdfs/papers-2020/TSCI190602056J.pdf.
[8] Jassim, E.I., Ahmed, F. (2020). Experimental assessment of Al₂O₃ and Cu nanofluids on the performance and heat leak of a double-pipe heat exchanger. Heat and Mass Transfer, 56: 1845-1858. https://doi.org/10.1007/s00231-020-02826-9
[9] Kannadhasan, V., Senthil Kumar, A., Vairamuthu, J., Nagarajan, R. (2022). Experimental research and CFD analysis on a double-pipe heat exchanger with CuO nanoparticles suspended in cold water. Journal of Thermal Analysis and Calorimetry, 147: 3831-3838. https://doi.org/10.1007/s10973-021-10804-4
[10] Jalili, B., Aghaee, N., Jalili, P., Ganji, D.D. (2022). Novel usage of the curved rectangular fin on the heat transfer of a double-pipe heat exchanger with a nanofluid. Case Studies in Thermal Engineering, 35: 102086. https://doi.org/10.1016/j.csite.2022.102086
[11] Nam, H.T., Lee, S., Kong, M., Lee, S. (2023). Numerical study of flow and heat transfer characteristics for Al₂O₃ nanofluid in a double-pipe helical coil heat exchanger. Micromachines, 14(12): 2219. https://doi.org/10.3390/mi14122219
[12] Bantan, R.A.R., Abu-Hamdeh, N.H., AlQemlas, T., Elamin, A.E.A.M.A. (2023). Heat transfer improvement of hybrid nanofluid with the use of twisted tapes within a heat exchanger. Alexandria Engineering Journal, 70: 673-684. https://doi.org/10.1016/j.aej.2023.03.016
[13] Azizi, Y., Bahramkhoo, M., Kazemi, A. (2023). Influence of non-uniform magnetic field on the thermal efficiency hydrodynamic characteristics of nanofluid in double pipe heat exchanger. Scientific Reports, 13: 407. https://doi.org/10.1038/s41598-022-26285-w
[14] Salim, A., Jihan, J.I., Das, B.K., Ahmed, M.R. (2024). Thermal performance investigation of N-shape double-pipe heat exchanger using Al₂O₃, TiO₂ and Fe₃O₄-based nanofluids. Energy Science & Engineering, 12(3): 1072-1090. https://doi.org/10.1002/ese3.1698
[15] Ajeeb, W., da Silva, R.R.T., Murshed, S.M.S. (2023). Experimental investigation of the heat-transfer performance of Al₂O₃ nanofluids in a compact plate heat exchanger. Applied Thermal Engineering, 218: 119321. https://doi.org/10.1016/j.applthermaleng.2022.119321
[16] Alhulaifi, A.S. (2024). Computational fluid dynamics heat transfer analysis of a double-pipe heat exchanger and flow characteristics using TiO₂ with water. Designs, 8(3): 39. https://doi.org/10.3390/designs8030039
[17] Kadhim, S.A., Askar, A.H., Saleh, A.A.M. (2024). An enhancement of double-pipe heat exchanger performance at a constant wall temperature using a nanofluid of iron oxide and refrigerant vapour. Journal of Thermal Engineering, 10(1): 78-87. https://doi.org/10.18186/thermal.1429191
[18] Kadhim, S.A., Hammoodi, K.A., Askar, A.H., Rashid, F.L., Abdul Wahhab, H.A. (2024). Feasibility review of using copper-oxide nanofluid to improve heat transfer in a double-tube heat exchanger. Results in Engineering, 24: 103227. https://doi.org/10.1016/j.rineng.2024.103227
[19] Gabir, M.M., Albayati, I.M., Hatami, M., Alkhafaji, D. (2024). An experimental investigation of the convective heat transfer augmentation in U-bend double pipe heat exchanger using water-MgO-Cmc fluid. Scientific Reports, 14: 12442. https://doi.org/10.1038/s41598-024-63043-6
[20] Abdelmagied, M. (2025). Numerical analysis of heat-transfer enhancement of Al₂O₃ and CuO-water nanofluids in annular curved tubes. International Journal of Air-Conditioning and Refrigeration, 33: 1. https://doi.org/10.1007/s44189-024-00066-8
[21] Mousazadeh, M., Sarmasti Emami, M.R., Afsari, K. (2025). Experimental assessment of MnCoFeO₄@H₂O nanofluid stability and its impact on heat-transfer enhancement in a double pipe heat exchanger. Results in Engineering, 27: 106022. https://doi.org/10.1016/j.rineng.2025.106022
[22] Canbolat, G. (2026). Numerical and theoretical investigation of thermal performance in double pipe heat exchanger using nanofluids with integrated CFD and ANN methods. Arabian Journal for Science and Engineering, 51: 21743-21760. https://doi.org/10.1007/s13369-026-11185-2
[23] Kern, D.Q. (1950). Process Heat Transfer. McGraw-Hill, New York.
[24] Aghayari, R., Maddah, H., Zarei, M., Dehghani, M., Ghanbari Kaskari Mahalle, S. (2014). Heat transfer of nanofluid in a double pipe heat exchanger. International Scholarly Research Notices, 2014(1): 736424. https://doi.org/10.1155/2014/736424
[25] Koo, J., Kleinstreuer, C. (2004). A new thermal conductivity model for nanofluids. Journal of Nanoparticle Research, 6: 577-588. https://doi.org/10.1007/s11051-004-3170-5
[26] Vajjha, R.S., Das, D.K. (2009). Experimental determination of thermal conductivity of three nanofluids and development of new correlations. International Journal of Heat and Mass Transfer, 52(21-22): 4675-4682. https://doi.org/10.1016/j.ijheatmasstransfer.2009.06.027
[27] Zheng, M.R., Han, D., Asif, F., Si, Z.T. (2020). Effect of Al₂O₃/water nanofluid on heat transfer of turbulent flow in the inner pipe of a double-pipe heat exchanger. Heat and Mass Transfer, 56: 1127-1140. https://doi.org/10.1007/s00231-019-02774-z
[28] Saha, G., Paul, M.C. (2014). Numerical analysis of the heat transfer behaviour of water based Al₂O₃ and TiO₂ nanofluids in a circular pipe under the turbulent flow condition. International Communications in Heat and Mass Transfer, 56: 96-108. https://doi.org/10.1016/j.icheatmasstransfer.2014.06.008
[29] Pak, B.C., Cho, Y.I. (1998). Hydrodynamic and heat transfer study of dispersed fluids with submicron metallic oxide particles. Experimental Heat Transfer, 11(2): 151-170. https://doi.org/10.1080/08916159808946559
[30] Xuan, Y.M., Roetzel, W. (2000). Conceptions for heat transfer correlation of nanofluids. International Journal of Heat and Mass Transfer, 43(19): 3701-3707. https://doi.org/10.1016/S0017-9310(99)00369-5
[31] Brinkman, H.C. (1952). The viscosity of concentrated suspensions and solutions. The Journal of Chemical Physics, 20(4): 571. https://doi.org/10.1063/1.1700493
[32] Maxwell, J.C. (1881). A Treatise on Electricity and Magnetism, 2nd ed. Clarendon Press, Oxford.
[33] Hamilton, R.L., Crosser, O.K. (1962). Thermal conductivity of heterogeneous two-component systems. Industrial & Engineering Chemistry Fundamentals, 1(3): 187-191. https://doi.org/10.1021/i160003a005
[34] Gnielinski, V. (1976). New equations for heat and mass transfer in turbulent pipe and channel flow. International Chemical Engineering, 16(2): 359-368.
[35] Dittus, F.W., Boelter, L.M.K. (1985). Heat transfer in automobile radiators of the tubular type. Reprinted in International Communications in Heat and Mass Transfer, 12(1): 3-22. https://doi.org/10.1016/0735-1933(85)90003-X
[36] Petukhov, B.S. (1970). Heat transfer and friction in turbulent pipe flow with variable physical properties. Advances in Heat Transfer, 6: 503-564. https://doi.org/10.1016/S0065-2717(08)70153-9
[37] Maïga, S.E.B., Nguyen, C.T., Galanis, N., Roy, G., Maré, T., Coqueux, M. (2006). Heat-transfer enhancement in turbulent tube flow using Al₂O₃ nanoparticle suspension. International Journal of Numerical Methods for Heat & Fluid Flow, 16(3): 275-292. https://doi.org/10.1108/09615530610649717
[38] Xuan, Y.M., Li, Q. (2003). Investigation on convective heat transfer and flow features of nanofluids. Journal of Heat Transfer, 125(1): 151-155. https://doi.org/10.1115/1.1532008
[39] Bejan, A. (1995). Entropy Generation Minimization: The Method of Thermodynamic Optimization of Finite-Size Systems and Finite-Time Processes. CRC Press, Boca Raton. https://doi.org/10.1201/9781482239171