Applied Thermal Engineering 50 (2013) 1543e1548

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Valorisation of low-temperature heat: Impact of the heat sink

on performance and economicsq
Oliver Buchin, Felix Ziegler*
Technische Universität Berlin, Institut für Energietechnik, KT 2, Marchstraße 18, D-10587 Berlin, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Low-grade heat is available everywhere; consequently, the valorisation of this heat seems to be attractive
Received 17 June 2011 in terms of economics. However, irrespective of the form of energy which is produced, any valorisation
Accepted 3 October 2011 comes along with the production of another stream of waste heat with even lower value. The dumping of
Available online 14 October 2011
this reject heat often turns out to be the issue which determines cost.
This presentation will elaborate on the influence of the heat sink temperature both on conversion
Keywords:
efficiency and cost. It first will give a frame on a very generic level. It is easy to reproduce the well-known
First cost
fact that the change in COP of a compression heat pump with heat sink or source temperatures is in the
Operating cost
Thermodynamics
order of some %/K. The same order of magnitude holds for all generic cycles with one important
Temperature exception: the influence of the heat sink temperature on the COP of a thermally driven cooling machine
Characteristic equation is about twice the impact of the other temperatures. In addition, simple equations to account for the cost
of heat exchange are presented. They show that heat pumps, be it work driven or heat driven, exhibit the
best efficiency-to-cost ratio.
In order to leave the generic level, a more detailed analysis is given for an absorption cooling system. It
is confirmed that the impact of the heat sink temperature on capacity and COP is significantly larger than
that of the other temperatures; in the nominal point a rise in heat sink temperature reduces the cooling
capacity by over 10%/K.
Finally, the influence of the humidity of the ambient air on performance is presented in a first order
approach, also.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction
Valorisation of low-temperature heat is a broad area. It covers
heat recovery by compression heat pumps, sorption heat pumps,
and heat transformers, as well as conversion into cold or
mechanical energy. The means to do so are abundant as well:
conventional conversion systems use thermodynamic mono-fluid
cycles (closed steam cycles, open steam cycles, or even gas
cycles), or dual-fluid cycles such as sorption cooling processes or
sorption power processes. Some options which are state of the art,
or technically feasible, or at least in discussion, are listed in Table 1.
The numbers (#1, 2, etc.) which are listed in Table 1, refer to
Fig. 1, in which the nature of the duty to be fulfilled by these cycles
is depicted on a temperature scale.
q A large part of this paper has been published before at the 2nd European
Conference on Polygeneration, March 30theApril 1st 2011, Tarragona, Spain.
* Corresponding author. Tel.: þ49 (0) 30 314 22387; fax: þ49 (0) 30 314 22253.
E-mail address: felix.ziegler@tu-berlin.de (F. Ziegler).
The first aim of this paper is to give an order to the said options
and then to elaborate on the impact of the heat sink, and that of the
other temperatures, too. It has to be distinguished between the
impact on power density, efficiency, and cost. This will be done
using generic equations. In a more technical approach, absorption
chillers will be discussed in detail.
2. Generic thermodynamic approach
2.1. Options for revalorisation
These options could be discussed in terms of exergy, but we
keep using energy and temperature as describing parameters.
First two definitions or clarifications are in order: low-
temperature heat is heat with a temperature above ambient (at
least two times the temperature gradient across a heat exchanger!).
The heat sink is defined by the ambient, as well. It may include
humidity, so it may be the dry bulb temperature or the wet bulb
temperature. In order not to complicate things, in the first part of
the paper we do not distinguish this.

1359-4311/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2011.10.002
1544 O. Buchin, F. Ziegler / Applied Thermal Engineering 50 (2013) 1543e1548

Table 1
Options for valorisation of low-temperature heat.

Mono-fluid Multi-fluid No fluid

Air cycle #1 Absorption cycle #2, 4, 5 Peltier cooler #1
Magnetocaloric cooler #1
Thermoelectric generator #3
Rankine cycle #3 Adsorption cycle #2, 4, 5
Steam jet cycle #2, 5 Chemical cycle #2, 4, 5
Stirling cycle #1, 3 Hybrid sorption cycle #1
Vapor compression cycle #1 Rankine sorption cycle #3
Vuilleumier cycle #2, 4, 5

2.2. Performance

Temperature of heat flow Table 2 gives the definition of efficiency or COP and an equation
#1: compression heat pump to determine it in the most simple way, which is to calculate the
work: T→∞ in out #2: heat-driven heat pump reversible limit and multiply this with a rough measure for ther-
#3: power cycle
#4: heat transformer
modynamic quality, g. This is done often although it is not really
#5: heat-driven refrigerator correct: irreversibilities, of course, change with temperature and,
drive: T4 in especially, with temperature differences. In order not to prolong
this paper we will stick to a simple numerical example, just for
upgraded: T3 out out out orientation. This example uses the temperatures from Table 3 and
a constant quality of g ¼ 0.5.
source: T2 in in in in in All these equations are well-known from textbooks (e.g. [1],) or
can be derived easily, so they will not be discussed here. The
sink: T1 out out out sensitivity of the efficiencies on the temperatures will be investi-
gated by using the derivatives. Table 4 gives the derivatives with
cold: T0 in respect to all the relevant temperatures, normalised with the
respective efficiency.
#1 #2 #3 #4 #5
From these equations it is obvious that, e.g., the impact of the
Fig. 1. Generic options to valorise heat of temperature T2.
source temperature, T2, on the heat pump COP is always somewhat
larger than that of the sink, T3, in the work driven case (#1) because
Fig. 1 schematically shows the thermodynamic options [1]; the 1 > T2/T3, whereas it is smaller in the heat driven case #2 because
only scale used is a temperature scale with temperature rising from (T4-T3)/(T4-T2)<T2/T3 under realistic circumstances. For a power
bottom to top. plant (#3) the impact of the heat sink, T1, always is somewhat larger
Each box denotes another generic conversion process. The than that of the heat source, T2 because 1 > T1/T2. In the case of the
words "in" or "out" are related to the energy flow on the respective heat transformer (#4) it is the other way round, again. This is
temperature level, entering or leaving the device. Box #1 depicts understood easily, as a shift in the intermediate temperature T2
a compression heat pump which upgrades heat from T2 to T3 with changes temperature lift and temperature thrust of the process at
input of mechanical work. Process #2 does the same upgrading, the same time. For the same reason, the impact of the heat sink in
albeit with using heat of temperature T4 as drive. It depicts a heat the case of refrigeration (#5) is the largest one.
driven heat pump. In order to quantify these findings, the derivatives according to
Box #3 stands for a power cycle which works between T2 and T1. Table 4 are plotted against the heat sink temperature for a set of
Box #4 represents a heat transformer cycle which upgrades heat other temperatures (see Table 3) in Figs. 2 to 6. The absolute value
from T2 to T3 with degrading part of the heat to the heat sink at T1. of the relative (normalised) derivatives is shown. The dashed lines
Box #5, finally, shows a heat driven refrigerator. So, summarising, display negative derivatives.
#1, #2, and #4 are heat pumps of different nature. The impact of the temperatures on the COP of a compression
In any case, the low-temperature heat source at T2 conveys heat heat pump (#1, Fig. 2) between T2 and T3 is the larger, the smaller
into the box. Energy flows out at a higher level (#1e4) or flows in at the temperature lift, (T3-T2), to be accomplished is. It is in the order
a lower level (#5). The focus of this paper is on the processes #3 to of some %/K for realistic temperatures. The difference between the
#5 because they have to reject heat to a heat sink at a temperature impact of the two temperatures is marginal.
T1. However, we will discuss processes #1 and #2 also shortly. For heat driven heat pumps (#2, Fig. 3), the impact of the
temperatures on the COP is in the same order of magnitude as for
compression heat pumps. However, the influence of the driving
Table 2
Performance of generic valorisation processes.
temperature T4 is the smallest and it varies slightly only. For
moderate temperatures, the impact of the sink T3 is the strongest,
Efficiency Example as discussed before.
#1 Work-driven heat pump Q T3 COP ¼ 4:9 For a power station which is operated by low-temperature heat
COP ¼ 3 ¼ g
W T3  T2
(#3, Fig. 4), the impact of the temperatures is reverse as compared
#2 Heat-driven heat pump Q T T  T2 COP ¼ 1:1
COP ¼ 3 ¼ g 3 4 to Fig. 2, naturally.
Q4 T4 T3  T2
#3 Power cycle W T  T1 h ¼ 0:06
h¼ ¼ g 2
Q2 T2 Table 3
#4 Heat transformer Q T T  T1 COP ¼ 0:29 Set temperatures according to Fig. 1.
COP ¼ 3 ¼ g 3 2
Q2 T2 T3  T1
#5 Heat-driven refrigerator Q T T  T1 COP ¼ 0:43 Temperature level T4 T3 T2 T1 T0
COP ¼ 0 ¼ g 0 2
Q2 T2 T1  T0 Value [ C] 180 120 80 35 5
 O. Buchin, F. Ziegler / Applied Thermal Engineering 50 (2013) 1543e1548 1545

Table 4
Relative change in COP or efficiency due to a change in one temperature.
   
vCOP  vh 
   
 vT  COP or vT  h; derivative with respect to.
i i

T4 T3 T2 T1 T0

T2 1
#1 
T3 ðT3  T2 Þ T3  T2
T2 T2 T4  T3
#2 
T4 ðT4  T2 Þ T3 ðT3  T2 Þ ðT3  T2 ÞðT4  T2 Þ
T1 1
#3 
T2 ðT2  T1 Þ T2  T1
T1 T1 T3  T2
#4  
T3 ðT3  T1 Þ T2 ðT2  T1 Þ ðT3  T1 ÞðT2  T1 Þ
T1 T2  T0 T1
#5 
T2 ðT2  T1 Þ ðT1  T0 ÞðT2  T1 Þ T0 ðT1  T0 Þ

for the other processes (be aware that the scale of the ordinate is
doubled!). So, it may be stated that waste heat driven chillers are
most sensitive to the respective temperatures. The impact of the
heat sink temperature T1 is by far the largest. We will elaborate on
these findings in a later chapter. Before we do so, we use the
equations from above for another short discussion with the focus
on first cost.

2.3. Impact on first cost

It is very well known that low efficiency drives operating cost.

Fig. 2. Relative change of the COP of a compression heat pump (#1) with
temperatures.
For the heat transformer (#4, Fig. 5) the relative impact of the
high temperature heat sink temperature T3 does not change with
temperature T2. Therefore, it is comparably small for small heat
source temperatures, but for high heat source temperatures it
becomes important. Then however, all derivatives are in the order
of 1%/K.
The most important result may be found in Fig. 6 for the
refrigerators (#5): all derivatives are about a factor of 2 larger than
However, low efficiency also drives first cost: the amount of heat
which has to be put through an energy conversion system, of
course, depends on its efficiency. Especially in processes which are
driven by low-temperature heat the cost for heat exchange
becomes a decisive issue [2]. In Table 5 equations for a specific cost
ratio s are given which is defined as the ratio of the overall heat
turnover to the useful energy (which may be heat or work):
P
jQ i j
s¼ (1)
Q USE
These equations may be combined with the equations for effi-
ciency in Table 2; then the derivatives in Table 4 can be applied. This
will not be exemplified here because it is a straightforward exercise.
It shall be sufficient to state that those temperatures which have
a large impact on efficiency will have a large impact on cost, also.
It is interesting to note that the relative heat turnover in the case
of the heat driven heat pump (#2) does not depend on the COP

Fig. 3. Relative change of the COP of a heat driven heat pump (#2) with temperatures. Fig. 4. Relative change of the efficiency of a power station (#3) with temperatures.
1546 O. Buchin, F. Ziegler / Applied Thermal Engineering 50 (2013) 1543e1548

Table 5
Specific cost ratio: relative heat turnover.

Relative heat turnover Example (see Table 2)

P
#1 Work-driven heat pump jQ i j 1 1.8
s¼ ¼ 2
Q COP
P 3
#2 Heat-driven heat pump jQ i j 2
s¼ ¼ 2
Q
P 3
#3 Power cycle jQ i j 2 32
s¼ ¼ 1
P
W h
#4 Heat transformer jQ i j 2 6.9
s¼ ¼
Q COP
P 3
#5 Heat-driven refrigerator jQ i j 2 6.7
s¼ ¼ 2þ
Q0 COP

Fig. 5. Relative change of the COP of a heat transformer (#4) with temperatures.
power, Q0, as well as the driving heat input, Q2, can be represented
as a linear function of a temperature functional, the characteristic
temperature function DDt, which, in turn, depends on the mean
temperatures of the external heat carriers, driving heat t2, cooling
water t1, and chilled water t0:

DDt ¼ at0  bt1 þ ct2 (2)

Q 0 ¼ S0 þ M0 DDt (3)

Q 2 ¼ S2 þ M2 DDt (4)
The benefit of this representation is the fact that the impact of
the temperatures is directly seen. As an example a single-effect LiBr
absorption chiller with a nominal cooling capacity of 10 kW shall be
used. In order to bring cost into the play the heat flows are nor-
malised by a first cost of 10,000V for this device. This, of course, is
not the real price of the chiller but it is in the right order of
magnitude. It renders a specific price of 1000V/kW of cooling
capacity. The cost-specific heat flows (W/V) then are given by:

q0 ¼ s0 þ m0 DDt (5)

Fig. 6. Relative change of the COP of a heat driven refrigerator (#5) with temperatures. q2 ¼ s2 þ m2 DDt (6)
The COP consequently is the ratio
because all input heat is turned into useful heat. In all other cases,
q0 s þ m0 DDt
the turnover of heat can only and will be reduced by increasing COP ¼ ¼ 0 (7)
efficiency. The numerical examples drastically show the problem of q2 s2 þ m2 DDt
producing power from low-temperature heat: as the efficiency of The coefficients are given in Table 6 [7]. The coefficient b gives
the power plant will be relatively small, the turnover of heat as the impact of the heat sink. It obviously is the largest one, and it is
compared to the power output will be large. The heat pumps (#1 about double the size as the others. The same order of magnitude
and 2), from this point of view, definitely show the best result. So resulted from the generic approach (Fig. 6) also. The respective
we can conclude at this stage that valorisation of waste heat by characteristic curves for cooling power, driving heat, and COP are
pumping it to a useful, higher temperature level may be the most shown in Fig. 7. The vertical line marks the design point of the
attractive option. This topic, however, will be covered by future chiller with a cooling capacity of 1W/V at a characteristic temper-
investigations. In this paper cooling from waste heat will be looked ature difference of about 22K.
at in more detail: cold is a more valuable product than heat and it is Any changes in any of the three temperatures (separately or
in the focus of today’s research. combined) may be seen as influencing cooling power or specific
cost, respectively. For instance, an increase in heat source temper-
ature t0 or driving heat temperature t2 will increase the charac-
3. Absorption chillers
teristic temperature function DDt according to equation 2.
Consequently, cooling power per investment, driving heat
3.1. Characteristic curves
requirement per investment, and COP will increase. This means
that for a given design the power increases, and for a given power
The discussion up to now, of course, suffers from the fact that
the specific features of the processes which are used do not show
up in the fundamental thermodynamic relationships. Therefore, in Table 6
order to check the validity, a more applied approach will be shown Coefficients for characteristics of an absorption chiller.
now. Coefficient a b c s0 m0 s2 m2
The performance of absorption chillers can be presented most Value 1.8 2.5 1 0.09W/V 0.042W/VK 0.2W/V 0.051W/VK
easily with using the characteristic equations [3e6]. The cooling
 O. Buchin, F. Ziegler / Applied Thermal Engineering 50 (2013) 1543e1548 1547

Fig. 7. Characteristic curves of a single-effect absorption chiller.

Fig. 8. Sensitivity of the specific cooling power on the temperatures.

the necessary investment decreases. The reverse influence is due to
the heat sink temperature t1.
3.2. Impact of temperatures on heat flows
In order to see the influence of the temperatures on heat flows
and the COP, again the derivatives have to be calculated. The results
for the heat flows are given in Table 7. The impact of the temper-
atures on the driving heat are stronger than on the cooling power.
Comparing the three temperatures, the influence of the heat sink
temperature t1 is the largest one.
The impact of the temperatures on the heat flows is almost
linear as stated before, but it is a non-linear impact on the COP.
Moreover, the relative impact is more interesting than the absolute
one. Therefore, the relative influence of each temperature on
relative cooling capacity and COP is plotted against the cost-specific
cooling power q0 in Figs. 8 and 9. Again, a negative value of the
derivative is indicated by a dashed line. The nominal point is
marked by a vertical line.
The impact of the temperatures on specific cooling capacity is in
the order of 5%/K (driving heat), 8%/K (chilled water), and over 10%/
K (heat sink) in the nominal point. It increases strongly when the
specific load is reduced. This happens in the case of part load, or in
the case of the characteristic temperature difference being small. In
this case, the required heat exchange area is relatively large, the
chiller will be expensive, and the sensitivity on the temperatures
will be high.
3.4. Impact of ambient humidity
We now also want to exemplify the influence of the ambient
conditions, namely dry bulb temperature and relative humidity,
separately. The absorption chiller is coupled to the ambient via the
heat sink, generally a wet or hybrid cooling tower. The cooling
water temperature t1 in this case is dependent on the characteris-
tics of the cooling tower also. We want to roughly model the system
dependence using the characteristic temperature functions for the
chiller. Ideal behaviour of the cooling tower is assumed: any heat
flow can be rejected as long as a driving temperature difference
across the exchanger exists. In order to take into account humidity,
the heat sink temperature is calculated as wet bulb temperature twb
of the ambient air. Furthermore the mean temperature difference
between humidified air and cooling water is set to 10 K. Equation
(2) in this case turns into
DDt ¼ at0  bðtwb ðta ; 4a Þ þ 10KÞ þ ct2 : (8)
Setting mean evaporator and generator temperatures and
ambient pressure constant (t0 ¼ 16.5  C, t2 ¼ 70.0  C, pa ¼ 1 bar) and
taking parameters for the chiller from Table 6 the normalized
cooling capacity Q0/10 kW, with Q0 according to equation (3), can
easily be calculated with respect to ambient temperature and
humidity (Fig. 10).

3.3. Impact of temperatures on COP

The impact on the COP is significantly less than the one on the
heat flows, as long as the specific load is not too small. This can be
seen in Fig. 7 as well as in Fig. 9, where the impact of the temper-
atures on the COP is plotted against the specific cooling load, again.
Once more, the derivative with respect to the heat sink tempera-
ture, t1, is the largest.

Table 7
Derivatives of the specific heat flows with respect to the temperatures.

i 0 1 2
vq0 a m0 0.076 -b m0 0.105 c m0 0.042
vti
vq2
a m2 0.092 -b m2 0.128 c m2 0.051
vti
Fig. 9. Sensitivity of the COP on the temperatures.
1548 O. Buchin, F. Ziegler / Applied Thermal Engineering 50 (2013) 1543e1548

using the approach of the characteristic functions. The impact of

the heat sink temperature on heat flows (capacity) and perfor-
mance (COP) is about twice as large as that of the other
temperatures.
Finally, a first attempt has been made to include the impact of
humidity via the wet bulb temperature into the calculus. This,
however has to be refined further by using a realistic model of the
cooling tower.
Summarising, the heat sink deserves more attention in the
design of processes which re-valorise low-grade heat. It can be
stated that the research which is dedicated to the field of heat
rejection does not match the importance within the area of energy
engineering.

Nomenclature

Fig. 10. Influence of ambient conditions on the normalized cooling capacity. a,b,c coefficients [-]
COP coefficient of performance [-]
g thermodynamic quality [-]
As expected, we see that for a given ambient temperature (dry
M coefficient [kW/K]
bulb) decreasing the relative humidity results in decreasing the
m coefficient [W/VK]
wet-bulb temperature and thus provoking higher cooling capac-
Q heat flow [kW]
ities. For a set cooling capacity, higher ambient temperatures can
q specific heat flow [W/V]
only be handled with lower relative humidities.
S coefficient [kW]
The system can be operated in free-cooling mode when wet
s coefficient [W/V]
bulb temperatures are lower than the required chilled water
T temperature (process) [K]
temperature, t0. This region is shown in the upper part of Fig. 10. In
t temperature (heat carrier) [K]
the respective example free-cooling corresponds to a normalized
W mechanical power [kW]
cooling capacity larger than 1.5.
h efficiency [-]
Hourly ambient conditions for a moderate climate (Berlin -
s relative heat turnover [-]
METEONORM Test Reference Year [8]) are plotted in Fig. 10 also. It
can be seen that the particular climatic conditions result in a wide
possible operating range of the chiller e in this example between Indices:
0.8 and 1.5 of the normalized cooling capacity at different relative 0.4 temperature levels
humidity, albeit at elevated dry bulb temperatures the operating a ambient
range starts to be restricted to part load. At maximum dry bulb
temperature (when the loads are highest) the normalized cooling
References
capacity will not surpass 0.8 as long as the relative humidity is
larger than 40%. [1] G. Alefeld, R. Radermacher, Heat Conversion Systems. CRC Press, Bocca Raton,
1993.
4. Conclusion [2] S. Köhler, F. Ziegler, Sorption cooling: best alternative for the use of low
temperature heat sources?, in: Proc. of Heat Powered Cycles 2006, Newcastle
upon Tyne, UK, 12th to 14th September(CD-ROM) (2006).
From the fundamental relationships which have been presented [3] T. Furukawa, et al., Study on characteristic temperatures of absorption heat
in this communication it can be concluded that the impact of the pumps, in: Proc. 20th Japan Heat Transfer Conf., June 1983 (1983), pp. 508e510
(in Japanese).
temperature and nature of the heat sink on performance and [4] T. Furukawa, T. Sonoda, Characteristics of H2O/LiBr absorption heat pumps for
economics in the field of valorisation of low-temperature heat is the temperature change of external Fluids, in: Proc. XVII Int. Congress of
predominant - except, of course, in the case of simple heat pumps. Refrigeration, Wien, 24.-29.8.1987 (1987).
[5] P. Riesch, J. Scharfe, F. Ziegler, J. Völkl, G. Alefeld, Part-Load Behavior of an
In all other cases the heat flow into this sink accounts for a large absorption heat transformer, in: Proc. 3rd Int. Symp. on the Large Scale
fraction of the energetic turnover. For the sake of efficiency, the Applications of Heat Pumps, Oxford. BHRA, The Fluid Engineering Centre,
driving temperature difference must be small which in turn Cranfield, Bedford, 1987, pp. 155e160.
[6] H.-M. Hellmann, C. Schweigler, F. Ziegler, A simple method for modelling the
necessitates large heat transfer areas. Heat pumps seem to be the
operating characteristics of absorption chillers, in: Proc. of Eurotherm Seminar
most cost-efficient devices for valorisation of low-temperature No.59, Nancy-Ville, France, 6-7 July (1998), pp. 219e226.
heat. [7] A. Kühn, F. Ziegler, Operational results of a 10 kW absorption chiller and
Of course a rising ambient or heat sink temperature in any case adaptation of the characteristic equation, in: Proc. Int. Conf. Solar Air Condi-
tioning, 6.-7. October 2005, Bad Staffelstein (2005).
reduces efficiency. This effect is especially large for heat driven [8] METEONORM software documentation, software version 6.0., METEOTEST,
cooling machines. These devices have been studied in more detail Fabrikstrasse 14, CH-3012 Bern, Switzerland.