Professional Documents
Culture Documents
Chantasiriwan2018 PDF
Chantasiriwan2018 PDF
Somchart Chantasiriwan
PII: S0260-8774(18)30076-1
DOI: 10.1016/j.jfoodeng.2018.02.021
Please cite this article as: Somchart Chantasiriwan, Online Determination of Heat Transfer
Coefficients in Sugar Juice Evaporation Process, Journal of Food Engineering (2018), doi: 10.1016
/j.jfoodeng.2018.02.021
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form.
Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
2 Process
3
4 Somchart Chantasiriwan
8 Abstract
10 raised to the boiling point, and the water content of the juice is reduced by evaporation.
11 The main process outputs are raw sugar and molasses. Heat transfer coefficients are
12 important parameters in the simulation of this process. Due to scaling, the values of heat
13 transfer coefficients decrease over time, and cleaning must be carried after a certain
15 recommended values of scaling rates are available in the literature, there are
17 evaporator station requires cleaning may be in error. In this paper, a method of online
18 determination of heat transfer coefficients and scaling rates using measured values of
19 process parameters is presented. Testing of this method indicates that it can produce
20 accurate values of heat transfer coefficients and scaling rates despite possible errors in
21 input data.
22
23 Key Words: Multiple-effect evaporator, Heat Exchanger, Mathematical model,
24 Parameter estimation
1
ACCEPTED MANUSCRIPT
25 1. Introduction
27 use of this tool can be found in the simulation of the sugar juice evaporation process, in
28 which a large proportion of the water content of incoming dilute sugar juice is removed,
29 and raw sugar and molasses are delivered as the final outputs. Three main components
30 of the process are multiple-effect evaporator, juice heaters, and crystallizer. Adams et al.
32 a control algorithm that could overcome difficulties arising from steam and juice
34 proposed by Burke (2014) was used to investigate the effects of operating conditions on
35 the Pareto optimal front, where juice flow rate is maximized and steam input to the
37 effect evaporator, which was used to show that the recovery of condensate flash vapor
39 considered the problem of optimum allocation of heating surface among the effects of
40 multiple-effect evaporator. Heluane et al. (2004) and Heluane et al. (2007) investigated
41 the problem of optimizing the sugar evaporation process through operation scheduling
43 procedure to optimize the design of the sugar juice evaporation system, which resulted
45 Chantasiriwan (2017) determined the optimum distributions of a fixed total juice heater
46 surface that result in the maximum rate of processed sugar juice and the minimum
2
ACCEPTED MANUSCRIPT
49 correlations of heat transfer coefficients for evaporators and juice heaters. Existing
50 models rely on correlations provided by Honig (1963), Smith and Taylor (1981), Guo et
51 al. (1983), Watson (1987), Van der Poel et al. (1998) for multiple-effect evaporator and
52 correlations provided by Wright (1981) and Hugot (1986) for juice heaters. It can be
53 shown that different correlations yield different values of heat transfer coefficients.
54 Correlations for evaporators are usually expressed in terms of juice concentration and
55 saturated steam temperature. Inspection of the derivation of one correlation carried out
56 by Wright (2008) revealed that there was significant scatter in the data from which the
57 correlation was derived. Rein (2007) pointed out that the uncertainty in heat transfer
58 coefficients might be due to factors not taken into account in correlations such as the
61 evaporators and juice heaters. Scaling is usually negligible in the steam side of the
62 evaporators or juice heaters. Scale deposits occur inside a tube containing flowing juice.
64 (2010) showed that the composition of scale varied from one sugar factory to another.
65 Solberg et al. (2006) reported a range of values of scaling rate in the final effect of the
70 scaling rates may not yield accurate results. More accurate values of heat transfer
71 coefficients and scaling rates for a specific sugar juice evaporation process should be
3
ACCEPTED MANUSCRIPT
73 transfer coefficients and scaling rates of heat exchangers in the sugar juice evaporation
75 temperatures, juice concentrations, and inlet juice flow rate, which are readily available
76 in most factories. Previous work by Solberg et al. (2006) presents a similar method for
77 determining heat transfer coefficients and scaling factors. However, Solberg et al.
78 (2006) restricted their analysis to only the multiple-effect evaporator by assuming that
79 there was no vapor bleeding from the multiple-effect evaporator. By contrast, the model
80 of the sugar juice evaporation system presented in this paper incorporates interaction
81 between the evaporators and juice heaters through mass and energy balances.
82
83 2. Sugar Juice Evaporation Process
85 evaporator system is shown in Fig. 1. A solid line represents steam and vapor, dashed
86 line represents sugar juice, and dotted line represents condensate. There are two juice
87 heaters. The primary juice heater (H2) receives diluted juice at the flow rate of mf,in from
88 the juice extraction process at temperature Th,2. The juice temperature is raised to Th,1
89 before the juice is passed through the secondary juice heater (H1). The juice temperature
90 at the exit of H1 is Th,0. The juice pressure at the exit of the secondary juice heater is
91 slightly above the atmospheric pressure. After passing through FC, dissolved gases in
92 the juice are removed, and its pressure is equal to the atmospheric pressure. Before
93 entering the first effect of the multiple-effect evaporator, the juice pressure is raised to
4
ACCEPTED MANUSCRIPT
96 p0. The thermal energy released by the condensation of the driving steam results in the
97 evaporation of water in sugar juice at a lower pressure (p1) in the first effect (E1). The
98 vapor leaving all effects except the last one (E5) is used to evaporate water in sugar
99 juice in the following effect. Because condensate from E1 is not contaminated with
100 dissolved solids, it is used as feed water for the boiler. Flash tanks are placed after E2,
101 E3, and E4. Flash tank F1 receives condensate at pressure p1 from E2 and H1, and
102 produces vapor and condensate at pressure p2. The vapor is used for evaporating water
103 in sugar juice in E3. F2 receives condensate at pressure p2 from F1, E3, and H2, and
104 produces vapor and condensate at pressure p3. The vapor is used for evaporating water
105 in sugar juice in E4. F3 receives condensate at pressure p3 from F2 and E4, and
106 produces vapor and condensate at pressure p4. The vapor is used for evaporating water
107 in sugar juice in E5. Condensates leaving F3 and E5 are collected in a storage tank.
108 The juice heaters require bled vapor from the evaporators. Vapors bled from the
109 first and second effects are used to increase the juice temperature in H1 and H2,
110 respectively. Additional vapor is bled from the first effect, and used in the crystallizer
111 (C) to evaporate most of the water content in the concentrated juice leaving the
112 evaporator E5. The outputs of the crystallizer are raw sugar and molasses.
113 The model of the sugar juice evaporation process consists of equations of mass
114 and energy balances, equations of steam and sugar juice properties, and correlations of
115 heat transfer coefficients. It can be seen from Fig. 1 that the mass balance equations are
116 m f ,1 m f , 0 mv ,1 mb ,1 ma (1)
117 m f , 2 m f ,1 mv , 2 mb , 2 (2)
118 m f ,3 m f , 2 mv ,3 (3)
5
ACCEPTED MANUSCRIPT
119 m f , 4 m f ,3 mv , 4 (4)
120 m f ,5 m f , 4 mv ,5 (5)
121 Equation (1) can be solved for mf,i in terms of mf,0, ma, mv,i, and mb,i.
i min i , 2
122 m f ,i m f , 0 ma mv , j m b, j (6)
j 1 j 1
124 1 1 mv,0hvl ,0 m f ,0 hfin,1 hfout,1 ma mv,1 mb,1 hv,1 hfout,1 (7)
126 1 1 mv, 2 mc,1 hvl , 2 m f , 2 hfin,3 hfout,3 mv,3 hv,3 hfout,3 (9)
127 1 1 mv,3 mc, 2 hvl ,3 m f ,3 hfin, 4 hfout, 4 mv, 4 hv, 4 hfout, 4 (10)
128 1 1 mv, 4 mc,3 hvl , 4 m f , 4 hfin,5 hfout,5 mv,5 hv,5 hfout,5 (11)
129 where hvl,i is the latent heat of evaporation at the saturation temperature Ti, hv,i is the
130 saturated steam enthalpy at Ti, and hf,i is the sugar juice enthalpy in effect i. It is
131 assumed that a fraction 1 of heat is lost in each vessel. Rein (2007) suggested that 1 =
132 0.015.
133 Figure 2 shows mass balances in F1, F2, and F3. The mass flow rate of vapor
134 from each flash tank (mc,i) is determined from the mass balance and energy balance of
139 where
6
ACCEPTED MANUSCRIPT
141 Juice temperature at the inlet to the first effect is below the boiling point.
142 Therefore, part of the heating surface in the first effect must be used for raising the juice
143 temperature to the boiling point before evaporation can occur. Juice temperature reaches
144 the boiling point at the exit of each effect. The effects of the concentration of dissolved
145 solids in juice and the hydrostatic pressure due to liquid level in the multiple-effect
146 evaporator result in the boiling temperature of sugar juice exceeding the saturated steam
3816.44 2x
148 T f 227.03 (16)
18.3036 ln7.5 p gH 2000 100 x
150 acceleration, is juice density, and H is liquid level in evaporator vessels. In this paper,
152 In addition to equations of mass and energy balances, there are the following
154
U1 A1 T0 T fout
,1
1 1 mv , 0 hvl , 0 (17)
155
U 2 A2 T1 T fout
,2
1 1 mv ,1hvl ,1 (18)
156
U 3 A3 T2 T fout
,3
1 1 mv , 2 mc ,1 hvl , 2 (19)
157
U 4 A4 T3 T fout
,4
1 1 mv ,3 mc , 2 hvl ,3 (20)
158
U 5 A5 T4 T fout
,5
1 1 mv , 4 mc ,3 hvl , 4 (21)
159 For the juice heaters, the requirement that the latent heat of condensation of the
160 bled vapor equals the increase in enthalpy of the juice in H1 and H2 yields
163 where cp,i is the average heat capacity of the juice between Th,i and Th,i1, and 2 is the
164 heat-loss fraction in the juice heaters, which is assumed to be 0.05. In addition, the heat
166 U h ,1 Ah ,1
Th,0 Th ,1
1 2 mb ,1hvl ,1 (24)
lnT1 Th ,1 T1 Th , 0
167 U h , 2 Ah , 2
T Th , 2
h ,1
1 2 mb , 2 hvl , 2 (25)
lnT2 Th , 2 T2 Th ,1
168 After leaving H1, the juice pressure (pin) is a little above the atmospheric pressure (pout).
169 The juice is allowed to flash in FC, resulting in a reduced mass flow rate (mf,0) that is
172 where Tin and Tout are saturation temperatures corresponding to pin and pout. As a
173 consequence, the juice concentration at the inlet to the first effect (x0) is related to the
174 juice concentration at the inlet to the primary juice heater (xin) as follows.
176 where i = 0 – 5.
177 Most of the water content of the syrup coming from the evaporators is removed
178 in the crystallizer, yielding raw sugar and molasses. The crystallizer is modeled as a
179 single-effect evaporator. Vapor bled from the first effect is used to evaporate most of
180 the remaining water in the syrup. Ideally, the amount of water to be evaporated is the
181 water content of the syrup. In practice, however, more water evaporation occurs due to
182 water addition and heat loss. Therefore, the ideal amount of vapor bleeding required for
8
ACCEPTED MANUSCRIPT
183 crystallization must be multiplied by a correction factor to obtained the actual amount of
184 vapor bleeding. It is assumed that the correction factor is 2.0. Therefore,
2m f ,5 1 x5 100 hvl ,5
185 ma (28)
hvl ,1
186
187 3. Determination of Heat Transfer Coefficients
188 Inspection of the above mathematical model reveals that there are 39 variables
189 (mf,in, mf,0 – mf,5, ma, mv,0 – mv,5, mb,1, mb,2, xin, x0 – x5, p0 – p5, Th,0 – Th,2, A1 – A5, Ah,1,
190 and Ah,2) and 26 equations [Eqs. (1) – (5), (7) – (11), (17) – (21), (22) – (28)]. Therefore,
191 the number of free parameters is 13. Some of the known parameters are 7 surface areas
192 of the evaporators and juice heaters (A1 – A5, Ah,1, and Ah,2). In a typical simulation, the
193 other 6 known parameters are the concentration of inlet sugar juice (xin), the pressure of
194 steam input to the multiple-effect evaporator (p0), the vapor pressure at the exit of the
195 last effect (p5), the juice temperature at the inlet of the primary juice heater (Th,2), the
196 juice temperature at the exit of the secondary juice heater (Th,0), and the mass flow rate
197 of processed juice (mf,in). Once these parameters are specified, the system of nonlinear
198 equations can be solved by iteration to find corresponding values of the remaining
199 variables.
200 It can be seen from Eqs. (17) – (21), (24), and (25) that U1 – U5, Uh,1, and Uh,2
201 can be determined if mv,0 – mv,5, mb,1, and mb,2 are known. Therefore, the problem of
202 determining heat transfer coefficients becomes the problem of determining these mass
203 flow rates using known values of mf,in, xin, x1 – x5, p0 – p5, Th,0 – Th,2, A1 – A5, Ah,1, and
204 Ah,2. Because mf,in is given, mf,0 can be determined from Eq. (26). Since mf,i can be
205 expressed in terms of mf,0, ma, mv,i, and mb,i, there are 9 unknowns in Eqs. (7) – (11),
9
ACCEPTED MANUSCRIPT
206 (22), and (23). Additional equations required for the determination of the unknowns are
207 obtained by eliminating mf,i from Eqs. (6) and (27). The resulting equations are
x
208 ma mv ,1 mb ,1 m f , 0 1 0 (29)
x1
x
209 ma mv ,1 mv , 2 mb ,1 mb , 2 m f , 0 1 0 (30)
x2
x
210 ma mv ,1 mv , 2 mv ,3 mb ,1 mb , 2 m f , 0 1 0 (31)
x3
x
211 ma mv ,1 mv , 2 mv ,3 mv , 4 mb ,1 mb , 2 m f , 0 1 0 (32)
x4
x
212 ma mv ,1 mv , 2 mv ,3 mv , 4 mv ,5 mb ,1 mb , 2 m f , 0 1 0 (33)
x5
213 Therefore, the system of equations consists of 9 unknowns and 12 equations. The
214 solution for ma, mv,0 – mv,5, mb,1, and mb,2 can be found by a least-squares method.
215
216 4. Results and Discussion
217 Consider a hypothetical sugar juice evaporation process. The surface area of the
218 first effect of the multiple-effect evaporator is 6000 m2, whereas the surface areas of the
219 other effects are 3000 m2. The surface areas of both the primary and secondary juice
220 heaters are 1000 m2. The values of heat transfer coefficients are 2.5, 2.5, 2.0, 1.5, and
221 0.7 W/m2.K for, respectively, the 1st, 2nd, 3rd, 4th, and 5th effects. The values of heat
222 transfer coefficients are 1.0 and 0.9 W/m2.K for the primary and secondary juice
223 heaters. By specifying the values of xin, p0, p5, Th,0, Th,2, and mf,in, simulated process data
10
ACCEPTED MANUSCRIPT
225 The system is assumed to be in operation for a period of 10 days. Due to scaling
226 or fouling effects, heat transfer coefficients are expected to decrease with time. It was
227 reported by Solberg et al. (2006) that the scaling rates in evaporator varied with time (t),
228 whereas the scaling rate in juice heaters was found by Watson and Wright (1985) to
229 vary with t0.8. Therefore, the variations with time of the heat transfer coefficients are
230 expressed as
1 1
231 n itn (34)
Ui Ui
1 1
232 n itn0.8 (35)
U h ,i U h ,i
233 where i and i are, respectively, scaling rates for the evaporators and juice heaters, and
234 t is time in hour. It is assumed that the scaling rates are 0.001 for the first 4 effects, and
235 0.008 for the last effect. The scaling rates for the primary and secondary juice heaters
237 Measurements of juice concentrations (xin, x1 – x5), steam and vapor pressures
238 (p0 – p5), juice temperatures (Th,0 – Th,2), juice mass flow rate (mf,in) are taken every hour
239 during the 10-day operation period. Figure 3(a) shows variations of juice concentrations
240 with time. It can be seen that only x5 decreases significantly with time, whereas the
241 other juice concentrations remain almost constant. Figure 3(b) shows variations of
242 pressures with time. Both p0 and p5 are kept constant in the simulation. The resulting
243 variations of the other pressures are quite small. Figure 3(c) shows that, due to effects of
244 scaling, the processed juice flow rate decreases monotonically with time. It should be
245 noted that variations of juice temperatures are not shown because all temperatures stay
11
ACCEPTED MANUSCRIPT
247 For the purpose of testing the proposed method, the measured values of xin, x1 –
248 x5, p0 – p5, Th,0 – Th,2, and mf,in are used to determine variations with time of U1 – U5,
249 Uh,1, and Uh,2. The measured values differ from the exact values due to limited precision
250 and accuracy. Limited precision results from the fact that a measured value can be
251 displayed to only a few decimal places. Limited accuracy is caused by a random error.
252 In Fig. 4, variations of U1 with time are presented in 3 plots corresponding to three
253 levels of precision and accuracy. The first plot corresponds to measurement data that
254 have the precision of 2 decimal places and the accuracy of 0.1 unit (which is
255 percentage for xin, x1 – x5, kPa for p0 – p5, C for Th,0 – Th,2, and kg/s for mf,in). The
256 second plot corresponds to measurement data that have the precision of 1 decimal place
257 and the accuracy of 0.5 unit. The third plot corresponds to measurement data that have
258 the precision of 1 decimal place and the accuracy of 1 unit. It can be seen that data
259 scatter increases as measurement data are less precise or less accurate. However, all
260 plots display the same trend. Regression analysis indicates that all three plots yield the
261 correct values of U1 and 1. Results for the other four effects are shown in Figs. 5 – 8.
262 Variations of Uh,1 and Uh,2 with t0.8 are shown in Figs. 9 and 10. Although data scatter
263 increases as measurement data become less precise or less accurate, correct values of
264 Uh,1, Uh,2, 1, and 2 can be recovered. It is interesting to note that data scatter in
265 computed values of Uh,1 is noticeably more than for Uh,2 because the smaller value of
266 the denominator in the log-mean temperature difference in Eq. (24) results in larger
268
269 5. Conclusion
12
ACCEPTED MANUSCRIPT
270 Accurate simulation of the sugar juice evaporation process requires accurate
271 values of heat transfer coefficients and scaling rates. Values provided in the literature
272 were obtained from experiments and test runs under specific conditions, which may be
273 different from conditions in a process. Using them in the process simulation may yield
274 inaccurate results. This paper presents a model of the sugar juice evaporation process, of
275 which components are evaporators, juice heaters, and crystallizer. The model can be
276 used to determine heat transfer coefficients and scaling rates of the evaporators and
277 juice heaters provided that process measurements of juice concentrations, juice
278 temperatures, evaporator effect pressures, and juice flow rate are available. Testing of
279 this method using simulated process measurements that have limited precision and
280 accuracy shows that accurate values of heat transfer coefficients and scaling rates can be
282
283 Nomenclature
291 T temperature, C
292 t time, h
13
ACCEPTED MANUSCRIPT
296
297 Greek symbols
302
303 Subscripts
312 v vapor
313 vl vapor-to-liquid
314
315 Superscripts
14
ACCEPTED MANUSCRIPT
319
320 References
321 Adams, G.J., Burke, B.J., Goodwin, G.C., Gravdahl, J.T., Peirce, R.D., & Rojas, A.J.
323 evaporators via nonlinear modelling and control. In Proceedings of the 17th
325 Korea.
326 Burke, B.J. (2014). Modelling and multi-objective optimisation of a sugar mill based
328 Federation of Automatic Control World Congress, Cape Town, South Africa.
331 Chantasiriwan, S. (2017). Distribution of juice heater surface for optimum performance
334 Ensinas A.V., Nebra S.A., Lozano M.A., & Serra L. (2007). Design of evaporation
335 systems and heaters networks in sugar cane factories using a thermoeconomic
337 Guo, S.Y., White, E.T., & Wright, P.G. (1983). Heat transfer coefficients for natural
340 Heluane, H., Colombo, M.A., Hernandez M.R., Sequeira, S.E., Graells, M., &
15
ACCEPTED MANUSCRIPT
343 1146.
344 Heluane, H., Colombo, M., Hernandez M.R., Graells, M., & Puigjaner, L. (2007).
347 Honig, P. (1963). Principles of Sugar Technology, Vol. III. Elsevier, Amsterdam.
348 Hugot, E. (1986). Handbook of Cane Sugar Engineering, 3rd ed. Elsevier, Amsterdam.
349 Jayes, W.E.G. (2004). Optimum distribution of heating surface in a multiple effect
351 Rackemann, D.W., Doherty, W.O.S., & East, C.P. (2010). Development of descriptor
352 tools for the characterization of Australian sugar mill evaporator scale.
353 Proceedings of the Australian Society of Sugar Cane Technologists, 27, 1-10.
355 Smith, I.A., & Taylor, L.A.W. (1981). Some data on heat transfer in multiple effect
358 Solberg, D., Rein, P., & Schlorke, D. (2006). Online evaporator heat transfer
360 Srivastava, D., Mohanty, B., & Bhargava, R. (2013). Modeling and simulation of MEE
361 system used in the sugar industry. Chemical Engineering Communications, 200,
362 1089-1101.
363 Van der Poel, P.W., Schiweck, H., & Schwartz, T. (1998). Sugar Technology: Beet and
16
ACCEPTED MANUSCRIPT
365 Watson, L.J. (1987). Heat transfer mechanisms in evaporators. Proceedings of the
367 Watson, L.J., & Wright, P.G. (1985). Fouling rates for tubular juice heaters.
369 Wright, P.G. (1981). Performance equations for horizontal tubular juice heaters.
371 Wright, P.G. (2008). Heat transfer coefficient correlations for Robert juice evaporators.
372 Proceedings of the Australian Society of Sugar Cane Technologists, 30, 547-
373 558.
17
ACCEPTED MANUSCRIPT
Figure Captions
1
Figure 1
E1 E2 E3 E4 E5
mv,0 mv,1 mv,2 mv,3 mv,4
cond
m c,1 c
enser mc,2 mc,3
FC F1 F2 F3
mf,0 mf,1 mf,2 mf,3 mf,4
mf,5
2
Figure 2
m +m m +m +m m +m
v,1 b,1 v,2 b,2 c,1 v,3 c,2
m m m
c,1 c,2 c,3
F1 F2 F3
m +m +m +m +m –m
v,1 b,1 v,2 b,2 v,3 c,3
m +m –m m +m +m +m –m
v,1 b,1 c,1 v,1 b,1 v,2 b,2 c,2
3
ACCEPTED MANUSCRIPT
Figure 3
80
(a)
60 x5
x4
xi (%)
40 x3
x2
x1
20
xin
0
0 50 100 150 200 250
t (h)
160
(b)
140
p1
120
100 p2
pi (kPa)
80 p3
60 p4
40
20 p5
0
0 50 100 150 200 250
t (h)
200
(c)
150
mf,in (kg/s)
100
50
0
0 50 100 150 200 250
t (h)
4
ACCEPTED MANUSCRIPT
Figure 4
0.8
0.6
1/U1 (m2 .K/kW)
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
0.8
0.6
1/U1 (m2 .K/kW)
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
0.8
0.6
1/U1 (m2 .K/kW)
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
5
ACCEPTED MANUSCRIPT
Figure 5
0.8
0.6
1/U2 (m2 .K/kW)
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
0.8
0.6
1/U2 (m2 .K/kW)
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
0.8
0.6
1/U2 (m2 .K/kW)
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
6
ACCEPTED MANUSCRIPT
Figure 6
0.8
0.6
1/U3 (m2 .K/kW)
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
0.8
1/U3 (m2 .K/kW)
0.6
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
0.8
1/U3 (m2 .K/kW)
0.6
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
7
ACCEPTED MANUSCRIPT
Figure 7
0.8
1/U4 (m2 .K/kW)
0.6
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
1.2
1
1/U4 (m2 .K/kW)
0.8
0.6
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
1.2
1
1/U4 (m2 .K/kW)
0.8
0.6
0.4
0.2
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
8
ACCEPTED MANUSCRIPT
Figure 8
3
1/U5 (m2 .K/kW)
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
3
1/U5 (m2 .K/kW)
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
3
1/U5 (m2 .K/kW)
0
0 24 48 72 96 120 144 168 192 216 240
t (h)
9
ACCEPTED MANUSCRIPT
Figure 9
1.4
1.2
1
1/Uh,1 (m2 .K/kW)
0.8
0.6
0.4
0.2
0
0 20 40 60 80
t0.8 (h0.8 )
1.4
1.2
1
1/Uh,1 (m2 .K/kW)
0.8
0.6
0.4
0.2
0
0 20 40 60 80
t0.8 (h0.8 )
1.6
1.4
1.2
1/Uh,1 (m2 .K/kW)
1
0.8
0.6
0.4
0.2
0
0 20 40 60 80
t0.8 (h0.8 )
10
ACCEPTED MANUSCRIPT
Figure 10
1.8
1.6
1.4
1/Uh,2 (m2 .K/kW)
1.2
1
0.8
0.6
0.4
0.2
0
0 20 40 60 80
t0.8 (h0.8 )
1.8
1.6
1.4
1/Uh,2 (m2 .K/kW)
1.2
1
0.8
0.6
0.4
0.2
0
0 20 40 60 80
t0.8 (h0.8 )
1.8
1.6
1.4
1/Uh,2 (m2 .K/kW)
1.2
1
0.8
0.6
0.4
0.2
0
0 20 40 60 80
t0.8 (h0.8 )
11
ACCEPTED MANUSCRIPT
Highlights