international conference on building energy, … · a multi-objective design optimisation strategy...

4th International Conference On Building Energy, Environment

Design optimisation of a double pass PV/T-solar air heater integrated with heat pipes

W. Fan1, G. Kokogiannakis1 and Z. Ma1

1 Sustainable Buildings Research Centre (SBRC),

University of Wollongong, Wollongong, NSW 2522, Australia

SUMMARYA multi-objective design optimisation strategy for a novel

design of hybrid photovoltaic thermal collector-solar air

heater (PVT-SAH) system integrated with heat pipes and fins

is developed to maximise its thermal and electrical

efficiencies. A dynamic model of the hybrid PVT-SAH system

is first developed. A sensitivity study using global sensitivity

analysis (GSA) is then implemented to identify the significant

parameters to the objective functions. Lastly, the

multi-objective optimisation problem is formulated using the

key parameters identified, objective functions defined,

genetic algorithm optimization technique and a

decision-making method. The results showed that the GSA

effectively reduced the optimisation size from 13 to 7. For the

PVT-SAH system with different lengths, both the thermal

efficiency and electrical efficiency were improved by 3.2-67.5%

and 9.9-25.2% respectively, in comparison to a baseline

design. The thermal efficiency of the optimised system was

ranged from 53.5 to 60% for different lengths of the

PVT-SAH system.

INTRODUCTIONHeat pipe is a heat transfer enhancement technology which

utilises evaporation and condensation of the inner fluid to

transfer a large amount of thermal energy. Although the heat

pipes have been widely applied in cooling of electrical

devices, there are a limited number of studies reported on

the integration of heat pipes with photovoltaic thermal

collectors (PVT) or solar air heater (SAH) systems (Gang et

al. 2011; Wu et al. 2011). In this study, a novel design of

hybrid PVT-SAH system integrated with heat pipes is

developed (Figure 1 and Figure 2) to increase the outlet air

temperature in order to drive rotary desiccant cooling

systems.

The hybrid system can be decomposed into two subsystems

for the convenience of analysis: a PVT module and a SAH

module (Figure 2). The PVT module consists of the glass

cover 1, PV panel, back plate, heat pipes and the lower air

channel. The SAH module includes the glass cover 2,

absorber plate, upper channel and the inserted longitude fins.

When operating, the ambient air is first circulated to the

lower channel where it is heated up by the heat generated

from the heat pipes. The heated air is then circulated to the

upper channel where the absorber plate of SAH will further

increase its temperature (Figure 2(b)). The PVT and SAH

modules are linked by a connection layer (using insulation

and adhesive materials) and the heat exchanges between

upper and lower channels could also occur through this

connection layer. The design of this PVT-SAH system

includes various design parameters such as geometric and

operational factors, and the effects of these parameters on

the thermal and electrical performance of the system are

usually nonlinear and complex. Improper selection of the

values of these parameters could decrease the performance

of such systems and further affect the feasibility of using

PVT-SAH to drive rotary desiccant cooling systems.

In this study, the performance of a hybrid PVT-SAH system

was optimised to maximise its useful thermal efficiency and

net electrical efficiency. An optimisation strategy using a mix

of global sensitivity analysis (GSA) and genetic algorithm

(GA) was used to facilitate the identification of the optimal

design.

Figure 1. Overview of the hybrid double pass PVT-SAH system

integrated with heat pipes

Figure 2. Section-view of (a) the hybrid PVT-SAH integrated with heat pipes;

(b) the upper and lower air flow channels

ISBN: 978-0-646-98213-7 COBEE2018-Paper277 page 834


METHOD

2.1. Outline of the multi-objective design optimisation

strategy

In this study, a design optimisation strategy for hybrid

PVT-SAH systems consisting of three steps was developed.

In the first step, a global sensitivity analysis was

implemented for dimension reduction to identify the most

influential design parameters. Secondly, based on the two

predefined objective functions and the performance model of

the PVT-SAH system, the key design parameters

determined through the sensitivity analysis were optimised

by using the multi-objective optimisation problem with

genetic algorithm as the optimisation technique. The

optimisation will result in a set of optimal Pareto solutions.

The TOPSIS decision-making method (Hwang and Yoon

2012) was used in the third step to determine the final

optimal solution from the set of Pareto solutions. TOPSIS is

based on the concept that the selected optimal point has the

shortest geographical distance from the positive ideal point

and the longest distance from the negative ideal point.

2.2. Global sensitivity analysis

Global sensitivity analysis (GSA) was used in this study to

evaluate the significance of each candidate design

parameter on the specified objective functions. GSA is a

variance based method and is able to measure the sensitivity

indexes for the models with up to 50 different input

parameters. Different from local sensitivity analysis methods,

the GSA can evaluate the integrated sensitivity of the design

parameter over its entire space. The GSA measures the total

contribution to the outcomes’ variation for each design

parameter, which includes the parameter’s main effect as

well as all the interactions involving that parameter. More

details of using GSA can be found in Cannavó (2012).

2.3. Objective functions

The optimisation in this study aims to maximise the useful

thermal efficiency and net electrical efficiency of the

PVT-SAH system. The calculation of the useful thermal

efficiency only took into account the outlet air with a

temperature greater than 60oC because at that temperature

the air can be used effectively for the regeneration process of

desiccant cooling systems. The calculation of the net

electrical efficiency considered the energy consumption of

the mechanical fan. The formulation of the two objective

functions is listed below:

Objective functions 1 (useful thermal efficiency):

𝑒𝑓𝑓𝑡ℎ = ∑ 𝑄𝑡ℎ𝑗𝑗=𝑁

𝑗=1∑ 𝐼𝑡

𝑗𝑗=𝑁𝑗=1⁄ (1)

In which 𝑄𝑡ℎ𝑗

= {∑ 𝐶𝑓

𝑗=𝑁𝑗=1 𝑚𝑓

𝑗 ∆𝑡 (𝑇𝑜𝑢𝑡

𝑗− 𝑇𝑖𝑛

𝑗), 𝑇𝑜𝑢𝑡

𝑗≥ 60℃

0, 𝑇𝑜𝑢𝑡𝑗

< 60℃(2)

where 𝐼𝑡 is the total radiation (𝐽) on the PVT-SAH system

during a time step; 𝑁 is the total time steps; ∆𝑡 is the length

of a time step (𝑠); 𝑗 is the 𝑗𝑡ℎ time step; 𝑇𝑜𝑢𝑡 and 𝑇𝑖𝑛 is

the outlet and inlet air temperatures (℃) respectively; 𝐶𝑓

and 𝑚𝑓 are the heat capacity (𝐽/𝑘𝑔 ∙ 𝐾) and mass flow rate

(𝑘𝑔/𝑠) of flowing air, respectively.

Objective function 2 (net electrical efficiency):

𝑒𝑓𝑓𝑛𝑒𝑡−𝑒𝑙𝑒 = ∑ (𝑄𝑒𝑙𝑒𝑗

− 𝑄𝑓𝑎𝑛𝑗

)𝑗=𝑁𝑗=1

∑ 𝐼𝑡𝑗𝑗=𝑁

𝑗=1⁄ (3)

where 𝑄𝑒𝑙𝑒 is the electricity ( 𝐽 ) generated by the PV

system; 𝑄𝑓𝑎𝑛 is the electricity (𝐽) consumed by the fan. The

friction coefficient used to calculate the 𝑄𝑓𝑎𝑛 can be found in

Bergman and Incropera (2011).

2.4. Design parameters and constraints

The number of the parameters considered for the design of

the hybrid PVT-SAH system is 13 (Table 1), which can be

grouped into three categories including geometrical, material

related and operational parameters. Defining the values or

ranges of the design parameters is crucial to preventing the

infeasible optimisation solutions. In this study, the values or

ranges of the design parameters were determined from

literature review and in case where there are no previous

published values, the ranges were set as wide as possible. A

summary of the considered design parameters and their

constraints is presented in Table 1. It is worth to note that the

solar properties of PV cells and absorber plate were fixed

and not optimised in this study. The PV cells considered in

this study were polycrystalline silicon with a solar absorption

of 0.9 and the reference average electrical efficiency of 18%.

Using solar selective coating, the absorber plate has the

solar absorption and emissivity of 0.95 and 0.05,

respectively.

Table 1 The specification of the candidate design parameters of the

hybrid PVT-SAH system

Candidate design parameters

Constraints Remarks

Material (Design 2006)

Steel; copper and aluminum

The material used to construct the absorber/fins and bottom plate which affects the input values of physical-thermal properties in the thermal model of PVT-SAH (such as thermal conductivity, heat capacity and density) .

PV covering factor 𝑟_𝑝𝑣

[0.1,0.9] The ratio of 𝑊_𝑃𝑉 to (𝑊_𝑃𝑉+𝑊_𝑆𝐴𝐻) as shown in Figure 2.

Number of fins 𝑁_𝑓𝑖𝑛

[1,50] The number of fins over the width of SAH (𝑊_𝑆𝐴𝐻) as shown in Figure 2.

Thickness of air gap 1 𝑡_𝑔𝑎𝑝1(m)

[0.01, 0.08] As shown in Figure 2

Thickness of air gap 2 𝑡_𝑔𝑎𝑝2(m)

[0.01, 0.08] As shown in Figure 2

Thickness of fins 𝑡_𝑓𝑖𝑛(m)

[0.001, 0.003]

As shown in Figure 2

Thickness of absorber plate 𝑡_𝑝(m)

[0.001, 0.003]

As shown in Figure 2

Thickness of glass cover 𝑡_𝑔𝑙𝑎𝑠𝑠

[0.001,0.005] As shown in Figure 2

Number of heat pipes (𝑁_ℎ𝑝)

[10, 220] The number of heat pipes installed along the length of PVT-SAH system.



Channel depth of lower channel 𝐻_𝑙𝑜𝑤𝑒𝑟(m) (Sun et al. 2010)

[0.01, 0.1] Channel depth of the PVT module.

Channel depth of the upper channel (𝐻_𝑢𝑝𝑝𝑒𝑟) (m)

[0.01, 0.1] The distance between the absorber plate and the bottom plate as shown in Figure 2.

Thickness of the connection layer (𝑡_𝑙𝑖𝑛𝑘) (m) (Board 2003)

[0, 0.1] The thickness of insulation layer between the SAH and PVT as shown in Figure 2. Glass wool was used in this study with a thermal conductivity of 0.04 W/m.KMass flow rate

(m_f) (𝑘𝑔 ℎ)⁄ [50, 700] -

2.5. Optimisation technique and decision-making method

Multi-objective genetic algorithm (MOGA) (Deb et al. 2000)

was used to optimise the significant design parameters that

were determined from the sensitivity analysis. The MOGA is

a specific class of genetic algorithm, which was developed

based on Pareto sorting technique. The optimisation will

result in a set of Pareto optimal solutions in a single run

which are non-dominated with respect to each other. The

non-dominated solution means the value of one objective

cannot be further improved unless sacrificing the other

objective. The advantage of the Pareto based multi-objective

optimisation is that it provides the insight of the trade-off

relationships among the several objectives and helps the

designer to make decisions. The MOGA therefore

overcomes many shortcomings of traditional aggregative

methods which combine all objectives into a weighted-sum

and only result in a single optimal solution.

The decision-making is essential for multi-objective design

optimisation problems to identify the final optimal design

from the set of Pareto fronts. The decision-making process is

generally performed depending on the engineering

experience or which objective function is more important to

the decision makers. In this study, the TOPSIS method is

used to determine the final optimal design. Two hypothetical

points, namely ideal point and worst point, were assumed to

assist in the decision-making process. The Pareto front

which has the shortest geometric distance from the ideal

point and the longest distance from the worst point was

determined to be the final optimal design. As the objective

functions are of different dimensions and scales, it is

necessary to normalize the objective functions before

implementing the TOPSIS method. The details about the

normalisation and implementation of TOPSIS method can be

found in Hwang et al. (1993).

2.6. Mathematical modelling of PVT-SAH system

A dynamic model of PVT-SAH system was first developed for

performance prediction. As the design of the present

PVT-SAH system can be considered as the combination of a

heat-pipe PVT module and a finned SAH module, the

prediction model of the whole system can be developed by

coupling a model of a standalone heat pipe PVT system

(Gang et al. 2011) with a model of a finned SAH (Fan et al.

2017). The nature of the coupling was to consider the heat

exchanges between the upper channel and lower channel

through the connection layer and to consider the heat and

mass transfer from the outlet of PVT to the inlet of SAH. The

inlet air temperature of the lower channel (PVT) was equal to

the ambient air temperature and its outlet temperature was

assigned to be the inlet air temperature of the upper channel

(SAH). Due to the mass continuity law, the mass flow rate in

the upper channel was equal to that in the lower channel.

Based on the above assumptions, the corresponding

modifications were made on the energy balance equations of

the existing models (Fan et al. 2017) and (Gang et al. 2011)

for the bottom plate of the SAH and the flowing air node in

the PVT while the remaining energy balance equations were

kept the same as those reported in the literature (Fan et al.

2017; Gang et al. 2011).

The energy balance for the bottom plate of SAH became:

𝐴𝑖 𝐶𝑏 𝑀𝑏 𝜕𝑇𝑏,𝑖 𝜕𝑡⁄ = 𝐴𝑐𝑠𝐾𝑓𝑖𝑛 (𝑇𝑓𝑖𝑛,𝑖 − 𝑇𝑏,𝑖) (𝐻𝑓𝑖𝑛 2⁄ )⁄ +

𝐴𝑖ℎ𝑟,𝑝−𝑏(𝑇𝑝,𝑖 − 𝑇𝑏,𝑖) − 𝐴𝑖ℎ𝑐,𝑏−𝑓2(𝑇𝑏,𝑖 − 𝑇𝑓2,𝑖) − 𝐴𝑖(𝑇𝑏,𝑖 −

𝑇𝑓1,𝑖)/𝑅𝑏𝑐,𝑓 (4)

where 𝐴 , 𝐶 , and 𝑀 are the areas (𝑚2 ), heat capacity

(𝐽/𝑘𝑔 ∙ 𝐾) and mass (𝑘𝑔/𝑚2); 𝑖 is the 𝑖𝑡ℎ control volume;

the subscripts 𝑏, 𝑓𝑖𝑛 , 𝑝 , 𝑓1 and 𝑓2 represent bottom

plate, fin, absorber plate, flowing air in the lower channel and

flowing air in upper channel respectively; 𝑇 stands for

temperature (℃) and 𝑡 for time step interval (s); ℎ is the

heat transfer coefficient; 𝑟 and 𝑐 in the subscript represent

radiation and convection heat transfer respectively; 𝑅𝑏𝑐,𝑓 is

the thermal resistance between the bottom plate and flowing

air in the lower channel. The left-hand side of the equation

includes the accumulated energy stored in the bottom plate,

and the right-hand side terms represent conduction heat

transfer from fins to the bottom plate, radiation heat transfer

from the absorber plate to bottom plate, convective heat

transfer from flowing air in the upper channel to bottom plate

and the combined heat transfer from bottom plate to the

flowing air in the lower channel.

Finally, the energy balance for the flowing air node in the

lower channel is as follows:

𝐶𝑓 𝜌𝑓 ∆𝑥 (𝐿𝑆𝐴𝐻 𝐻𝑢𝑝𝑝𝑒𝑟) 𝜕𝑇𝑓1,𝑖 𝜕𝑡⁄ +

𝐶𝑓 (𝑚_𝑓 3600)⁄ ∆𝑥 𝜕𝑇𝑓1,𝑖 𝜕𝑥⁄ = 𝜋𝑑𝑐𝑜,𝑜𝑢𝑡𝐿𝑐𝑜ℎ𝑐,𝑐𝑜−𝑓1(𝑇𝑐𝑜,𝑖 −

𝑇𝑓1,𝑖) + 𝐴𝑖(𝑇𝑏,𝑖 − 𝑇𝑓1,𝑖)/𝑅𝑏𝑐,𝑓 + 𝐴𝑖(𝑇𝑎𝑚𝑏 − 𝑇𝑓1,𝑖)/𝑅𝑓,𝑎 (5)

where 𝜌𝑓 is the air density (𝑘𝑔/𝑚3); ∆x is the length of a

control volume along the PVT air channel (𝑚); 𝐿𝑆𝐴𝐻 𝐻𝑢𝑝𝑝𝑒𝑟

is the cross-section area of the upper channel (𝑚2); 𝑚𝑓 is

the mass flow rate (𝑘𝑔/ℎ); 𝐿𝑐𝑜 is the length of condenser

(m); 𝑑𝑐𝑜,𝑜𝑢𝑡 is the diameter of outside surface of the

condenser ( 𝑚 ); ℎ𝑐,𝑐𝑜−𝑓1 is the convection heat transfer

coefficient (W/(𝑚2 ∙ 𝐾)) in the lower channel; 𝑅𝑓,𝑎 is the

thermal resistance ((𝑚2 ∙ 𝐾)/𝑊) of the bottom surface layer

of the lower channel; 𝑇𝑎 is the ambient air temperature. The

left-hand side terms in Eq. (5) represent energy storage in

the control volume for the flowing air node and the

temperature gradient along the flow direction, while the



right-hand side terms are the convection heat transfer from

the condenser to the flowing air, the convection heat transfer

from the duct air to the bottom plate of SAH, and the

convection and conduction heat transfer from flowing air in

the channel to the surrounding ambient air.

The energy balance equations of all construction

components of the PVT module and SAH module were

interlinked in nature and were arranged into a matrix form so

that the temperature of construction components can be

simultaneously solved. The detailed descriptions of the

methods for equation handling and model solution were

given by Fan et al. (2017). With the time varying inputs of the

boundary conditions (i.e. weather conditions, mass flow rate),

the model can predict the temperatures of outlet air and the

construction components of the PVT-SAH system over time

and length. The outlet air temperature and PV working

temperature can be respectively used to calculate the useful

thermal efficiency and net electrical efficiency as shown in

Eqs. (1) and (3).

RESULTS AND DISCUSSIONAll the simulations in this study were conducted under the

typical summer week conditions of Darwin city in Australia as

shown in Figure 3. The climate in Darwin has frequent high

temperatures and high level of humidity during summer

period which provides great potential for the application of

PVT-SAH integrated desiccant cooling systems. The

proposed PVT-SAH system was assumed to be mounted on

the roof of a typical Australian residential building with the

average roof tilt angle of 23.6o. The performance of heat pipe

is influenced by its inclination angle which was equal to the

roof tilt angle in this study with the condenser at higher

position. To accommodate different lengths of the roof, the

design optimisation was performed for different lengths of the

PVT-SAH systems (i.e. 6, 9, 12, 16, 19, 22 m), respectively.

As the objective functions are the efficiencies in terms of the

thermal and electrical energy rather than absolute amount of

energy, the unit width of PVT-SAH was considered in all

simulations.

3.1. Results from the sensitivity analysis

Given the constraints of the design parameters provided in

Table 1, the sensitivity index was evaluated for each design

parameter with respect of the useful thermal efficiency and

net electrical efficiency respectively. From Figure 4, it can be

seen that the material parameter did not significantly

influence the outputs of useful thermal efficiency and net

electrical efficiency. In addition, the joint contributions of the

design parameters such as the thickness of glass cover

(t_glass), air gap1/2 (t_gap1 and t_gap2), absorber plate

(t_p) and fins (t_fin) to the objective functions were small in

comparison to the other parameters. Hence, the parameters

mentioned above were grouped as low sensitivity

parameters and were not further considered in the following

optimisation. The level of significance for each key design

parameter on the useful thermal efficiency and net electrical

efficiency is different. The mass flow rate (m_f), PV covering

ratio (r_pv) and the number of heat pipes (N_hp) were the

top three most sensitive parameters to the useful thermal

efficiency. On the other hand, the parameters that contribute

most to the improvement of the net electrical efficiency were

the lower channel depth (H_lower) and upper channel

(H_upper) depth, as well as the mass flow rate (m_f). It is

worthwhile to note that the sum of the total sensitivity

indexes of each objective function is more than one,

indicating the existence of the interactions among some of

the parameters.

Figure 5 represents the variations of the useful thermal

efficiency and net electrical efficiency with the change of

each key design parameter while keeping the other

parameters constant and by using the base design data

provided in Table 2. The base design in Table 2 was

determined to ensure that the PVT-SAH has the

performance at an average level. Figure 5 (c) and (e)

Figure 3. The weather conditions of a summer week of Darwin City,

Australia.

(a) Useful thermal efficiency

(b) Net electrical efficiency

Figure 4. The sensitivity indices of (a) useful thermal efficiency,

(b) net electrical efficiency for different design parameters.



showed the conflicting effect that the channel depth has on

the outcomes of η_th and η_ele. The smaller channel depth

was advantageous for increasing the thermal performance,

but more electricity was consumed by the fan to drive the

flowing air. Figure 5 (f) showed that there was an optimal

mass flow rate for both η_th and η_ele. When mass flow

rate was greater than a specific value, the obtained useful

thermal energy (with temperature over 60oC) was reduced

and the fan electricity consumption was magnified, which

resulted in a lower η_th and a lower η_ele. From Figure 5

(g), it can be seen that a thicker thermal insulation between

upper and lower channels can contribute to increasing the η

_th and η_ele. The reason is that the sufficient insulation can

prevent the heat transfer from the upper channel to the lower

channel, resulting in a lower temperature of the flowing air in

the lower channel and thus increased the temperature

difference of the convection heat transfer between flowing air

and the condenser. This illustrates that an air gap between

the upper and lower channels could be another alternative in

addition to using solid insulation material as an air gap can

also provide sufficiently high thermal resistance.

Table 2. The reference design of the hybrid PVT-SAH

N_hp r_pv H_lower

(m)

N_fin H_upper

(m)

t_link

(m)

m_f

(𝑘𝑔/ℎ)

90 0.5 0.05 18 0.05 0.02 200

3.2. Results from the multi-objective optimisation

The multi-objective optimisation problem of the PVT-SAH

system with different lengths was solved by using

multi-objective genetic algorithm optimizer and the

decision-making method. For each optimisation case, the GA

optimizer will result in a set of Pareto Fronts solutions. Using

the PVT-SAH with a length of 6 meters as an example, the

obtained optimal Pareto Fronts is shown in Figure 6 (a).

From Figure 6 (a), it is observed that the ideal maximum

useful thermal efficiency (Pont A) and net electrical efficiency

(Point B) cannot be achieved simultaneously and the results

of the two objective functions were in a conflicting

relationship. This means that the final optimal solution is a

tradeoff between the two objective functions. Theoretically,

all the solution points on the Pareto Fronts curve are optimal

and the selection of the final candidate solution is usually

dependent on which objective is of more important to the

designers. The TOPSIS method was used to locate the final

optimal solution in this study. Figure 6 (b) presents the

normalized Pareto Fronts curve, on which an ideal positive

point and worst point were assumed to assist in the selection

of the final optimal point. The candidate point which had the

longest geographical distance from the ideal worst point and

the shortest distance from the ideal positive point was

determined to be the final optimal solution. Using the same

method, the optimum key design parameters and the

resultant performance for the PVT-SAH system with different

lengths were obtained and are summarised in Table 3. The

optimal useful thermal efficiency and net electrical efficiency

were in the ranges of 53.5-60% and 7.61-9.76% respectively

for different lengths of the PVT-SAH system. It is found that

the optimal number of the heat tubes and the mass flow rate

increased with the increase in the length of PVT-SAH system.

The thicker insulation layer between the SAH and PVT was

preferred for all optimisation cases. A further comparison of

the useful thermal efficiency and net electrical efficiency

between the obtained optimal design (Table 3) and the

baseline design (Table 2) was conducted and the result is

shown in Figure 7. It can be seen that the performance of the

optimal design was significantly higher than the baseline

design with η_th increased by 3.2-67.5% and η_ele

increased by 9.9-25.2% in comparison to the baseline design

for different lengths of the PVT-SAH system. These results

demonstrated the effectiveness of the multi-objective design

Figure 6. (a) optimal Pareto fronts without normalisation;

(b) normalized Pareto fronts Pareto fronts.

(a)

(b)

Figure 5. The variations of useful thermal efficiency and net electrical

efficiency verse the change of (a) N_hp, (b) r_pv, (c) H_lower, (d)

N_fin, (e) H_upper, (f) m_f, (g) t_link.



optimisation strategy to optimise the hybrid PVT-SAH

system.

Table 3 The optimal design of PVT-SAH and the resulted

performance for different lengths of PVT-SAH system

Length of PVT-SAH system (m)

6 10 13 16 19 22 Optimal performance

η_th (%) 53.5 58.2 60 58.9 59.1 56.3

η_ele (%) 9.76 8.93 8.36 8.11 7.61 8.31

Optimal design of parameters

N_hp 79 93 145 176 161 202

r_pv 0.66 0.56 0.52 0.52 0.47 0.59

H_lower(m)

(m)

0.06 0.06

5

0.08 0.052 0.05

4

0.058

N_fin 13 16 23 31 25 16

H_upper (m)

(m)(m)

0.03

5

0.04 0.05

2

0.088 0.07

1

0.089

t_link (m) 0.1 0.1 0.1 0.1 0.1 0.1

m_f (kg/h) 155 285 370 430 472 596

4. ConclusionsThis study aimed at optimising the useful thermal efficiency

and net electrical efficiency of a heat pipe integrated

PVT-SAH system. For this purpose, a multi-objective design

optimisation strategy using genetic algorithm in combination

with a global sensitivity analysis (GSA) was proposed.

The main conclusions derived from this study are:

• The GSA was effective for dimension reduction. Seven

design parameters were finally determined to be

significant to the objective functions and were

considered in the optimization; • The construction material and the thickness of

construction components (except the insulation layer

between the upper and lower channel) did not have a

significant effect on the thermal and electrical

performance. This indicated that the light-weight

PVT-SAH and using the materials with a lower price

like steel are favored to reduce the upfront costs for

constructing the PVT-SAH system without lessening

the performance, although there are other factors that

could also affect such decisions (e.g. resistance to

corrosion);

• The optimal useful thermal efficiency and net electrical

efficiency cannot be obtained simultaneously. A

decision-making strategy was needed to choose a

compromised design as the optimal design;

• The optimised values of the useful thermal efficiency

and net electrical efficiency were in the ranges of

53.5-60% and 7.61-9.76%, respectively.

• The comparison study demonstrated that

multi-objective optimisation strategy was technically

feasible and could solve the optimisation problem to

maximise both thermal and electrical performance of

hybrid double pass heat pipe PVT- finned SAH

systems.

Acknowledgements The authors would like to thank China Scholarship Council

and University of Wollongong for supporting this study.

5. ReferenceBergman, T.L., Incropera, F.P., 2011. Fundamentals of heat and

mass transfer. John Wiley & Sons. Board, A.B.C., 2003. Building code of Australia. CCH Australia. Cannavó, F., 2012. Sensitivity analysis for volcanic source modeling

quality assessment and model selection. Computers &

Geosciences 44, 52-59. Deb, K., Agrawal, S., Pratap, A., Meyarivan, T., 2000. A fast elitist

non-dominated sorting genetic algorithm for multi-objective

optimization: NSGA-II, International Conference on Parallel

Problem Solving From Nature. Springer, pp. 849-858. Design, C.G.A.E., 2006. The Chartered Institution of Building

Services Engineers. London. Fan, W., Kokogiannakis, G., Ma, Z., Cooper, P., 2017. Development

of a dynamic model for a hybrid photovoltaic thermal collector –

Solar air heater with fins. Renewable Energy 101, 816-834. Gang, P., Huide, F., Tao, Z., Jie, J., 2011. A numerical and

experimental study on a heat pipe PV/T system. Solar energy

85(5), 911-921. Hwang, C.L., Lai, Y.J., Liu, T.Y., 1993. A new approach for multiple

objective decision making. Computers & operations research

20(8), 889-899. Hwang, C.L., Yoon, K., 2012. Multiple attribute decision making:

methods and applications a state-of-the-art survey. Springer

Science & Business Media. Sun, W., Ji, J., He, W., 2010. Influence of channel depth on the

performance of solar air heaters. Energy 35(10), 4201-4207. Wu, S.Y., Zhang, Q.L., Xiao, L., Guo, F.H., 2011. A heat pipe

photovoltaic/thermal (PV/T) hybrid system and its performance

evaluation. Energy and buildings 43(12), 3558-3567.

Figure 7. Performance comparisons between the optimal design and

baseline design for different lengths of the hybrid PVT-SAH system.


international conference on building energy, … · a multi-objective design optimisation strategy...

Documents