Copyright © 2011 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol. 5, N. 2 Special Issue on Heat Transfer
301
TABLE II
M
EAN
R
ELATIVE
A
ND
A
BSOLUTE
E
RRORS
O
F
N
EURAL
N
ETWORK
M
ODEL
Variables Feedforward Neural
Network model Radial Basis Neural
Network Model Mean
Relative Error
Mean Absolute
Error Mean
Relative Error
Mean Absolute
Error T
2
0.0184 °C 25.35
0.0108 °C 9.68
RH
2
0.3743 1.152 0.3935 1.227 T
9
0.0194 °C 3.39
0.0178 °C 3.124
RH
9
0.2834 0.763
0.366 0.966
II. Design Conditions
The design of the desiccant cooling system is controlled by many operating conditions.
Referring to Figures 1 and 2, the following parameters may be used as a basis for designing the system: ambient
conditions, inside room conditions, regeneration air temperature before the dehumidifier, supply and return
air flow rates, and design sensible and latent cooling loads:
Ambient conditions: Ambient conditions are based on ARI Standard 1060 2005. These values are T
1
DB=35 °C, T
1
WB=24 °C.
Inside room conditions: Recommended standard design conditions for a residential air conditioner are
based on ARI Standard 1060 2005 and the values are T
5
DB=26.7 °C, RH
5
= 50. Regeneration air temperature: Hence, fixing the final
state of air for the regeneration process would give a better picture of the influence of this constraint on the
cycle. In this study, this was fixed at various temperatures 75, 90, 100
°C. These temperatures are also proven acceptable
according to the studies performed on a rotary dehumidifier.
Supply and regeneration air flow rates: Modelling of the dehumidifier was performed taking into account that
the system runs with volume air flow ratio between regeneration and supply side r, equal or less than 1
one.
Sensible and latent cooling loads: Based on the potential application of the system on different building
uses the Sensible Heat Factor, hence the sensible to total heat ratio was varied.
III. Ventilation Cycle Analysis
Referring to Figure 1, warm and wet ambient air is introduced in the process air at State 1.
This process air is dried, while it passes over the desiccant, resulting in hot, dry air as it exits the
dehumidifier of State 2. This increase in temperature is due to the release of
heat of condensation of the water vapour when moisture is removed by the desiccant material.
The air of State 2 is then heat exchanged with room air that is adiabatically humidified to State 6 to create air
at State 3. This air is then humidified to State 4.
For the regeneration air stream, indoor air at State 5 is humidified to State 6.
This air is then heat exchanged with the process air stream at State 2 to produce air at State 7.
This air is then heated to the regeneration State 8. After regenerating the desiccant, air at State 9 is then
exhausted back to the outdoors. Following, the ventilation cycle for the three
aforementioned models of the desiccant wheel is presented in the psychrometric chart bellow, developed
for the Lab of Applied Thermodynamics of 40 kW cooling capacity and 30 of latent loads.
Moreover, all the components are considered as ideal. From the psychrometric chart bellow it could be
noticed that there is an increase in dry bulb temperature of State 2 due to the release of heat of condensation of
the water vapour when moisture is removed by the desiccant material.
This release of heat after dehumidification is greater in Beccali’s and NNMDCS’ models interpolation of
experimental results, resulting also in greater Coefficient of Performance for the whole cycle.
The non-linear analogy method of Bank’s overestimates the dehumidification capability of the
desiccant wheel.
Fig. 4. Psychrometric chart of the ideal ventilation cycle using neural network’s, Beccali’s and Banks’ models
IV. Effects of Components Characteristics
