nd
im
*, T
ent,
2004
line 1
ioned
ould
s po
A case study of a Swedish kraft pulp mill showed that 5–6 MW excess heat with a temperature above 90 �C could be released if
the system were rebuilt. Two different alternatives for using the excess heat have been investigated.
� 2005 Elsevier Ltd. All rights reserved.
closed water loops and greenhouse gas abatement poli-
cies. This could lead to new opportunities for energy
fined by the hot streams that produce process water or
are cooled in coolers and the cold streams that represent
opportunities for energy saving; see e.g. [3–5]. Combin-
ing energy and water pinch to include possible synergis-
tic effects has also been studied, e.g. [6].
The definition of usable excess heat in this work
corresponds to the one stated by Wising [2]:
.
* Corresponding author. Tel.: +46 31 772 30 22; fax: +46 31 82 19
28.
E-mail address: roger.nordman@chalmers.se (R. Nordman).
Applied Thermal Engineering 2
1359-4311/$ - see front matter � 2005 Elsevier Ltd. All rights reserved
savings. The fact that the price of energy has increased
substantially in recent years makes energy-saving pro-
jects ever more attractive [1,2].
In existing mills, warm and hot water is often over-
produced and both warm and hot water tanks are
commonly overflowed to a cooling tower, or drained
away. Instead of producing this surplus, the heat (cool-
ing demand) could be used for other purposes [1,2].
Since the main objective for the pulp mill is to produce
high-quality pulp, a change in the heat recovery system
the process water heat demand. The method shows how
much energy at a high temperature it is theoretically and
practically possible to release through a new design
method for the HWWS.
2. Theory
Many process integration (PI) studies in the pulp and
paper industry have been done, mostly to investigate the
Keywords: Pulp mill; Heat recovery, Pinch analysis; Optimization; Hot and warm water system
1. Background
The pulp and paper industry faces new challenges
and new operating conditions. Examples are stricter
environmental regulations that increase the need for
should not affect the process in a negative way. There-
fore a new method has been developed to maximize
the amount of usable excess heat (hereafter referred to
as Qxs) in the hot and warm water system, HWWS (sec-
ondary heat system) in a pulp mill. The HWWS is de-
Design of kraft pulp mill hot a
method that max
Roger Nordman
Heat and Power Technology, Department of Energy and Environm
Received 28 December
Available on
Abstract
Many pulp mill hot and warm water systems are overdimens
overproduction could be used for other applications, if heat c
method to release as much excess heat of a high temperature a
doi:10.1016/j.applthermaleng.2005.06.001
hore Berntsson
Chalmers University of Technology, SE 41296 Go¨teborg, Sweden
; accepted 3 June 2005
9 July 2005
and produce more hot and warm water than demanded. This
be released at high enough temperatures. In this paper a new
ssible in the secondary heat system has been developed.
warm water systems—A new
izes excess heat
www.elsevier.com/locate/apthermeng
6 (2006) 363–373
Nomenclature
AHX heat exchanger area (m
2)
AEvap total evaporator area (m
2)
Cel electricity price (US$/MW h)
Cfuel fuel price (US$/MW h)
CDH district heat price (US$/MW h)
Clost lost electricity production (kW)
h00 saturated steam enthalpy (kJ/kg)
h 0 saturated liquid enthalpy (kJ/kg)
n number of evaporator effects used by excess
heat
N total number of evaporator effects
Qxs usable excess heat (kW)
QC cooling demand (kW)
QH heat demand (kW)
Qtot total evaporation heat demand (kW)
Qfuel fuel demand (kW)
top time of operation (h/yr)
VTank tank volume (m
3)
C ratio of saved live steam to excess heat (–)
DTmin minimum temperature difference (K)
gb boiler efficiency (–)
gT turbine efficiency (–)
gm+g mechanical and generation efficiency (–)
CC composite curve
CTMP chemi-thermo mechanical pulp
FOB free on board
NAS neat annual savings
HEN heat exchanger network
HWWS hot and warm water system (also referred to
as ‘‘secondary heat system’’)
TTL tank temperature level
364 R. Nordman, T. Berntsson / Applied Thermal Engineering 26 (2006) 363–373
‘‘The definition of usable excess heat is heat sources with
a high temperature, typically above 80 �C but still below
the pinch temperature, which is not needed for warm
and hot water heating. The expression �usable excess
heat� implies that it can be used in the process to reduce
energy demand but by pinch analysis definition there
cannot be any �excess heat� below the pinch temperature.
However if the evaporation plant is regarded as a flexi-
ble unit that is allowed to be redesigned, the heat source
below pinch temperature can be used for evaporation,
substituting live steam. Heat could also be exported to
other processes, where it could be used, hence the
expression usable excess heat.’’
Q (kW)
Excess heat
T
(°
C
)
a
Fig. 1. (a) Composite curves show how the temperature of the cooling deman
until reaching the HWWS DTmin yield the maximum usable Qxs.
3. Process description
The kraft process consists of two main parts: the fibre
line where pulp is produced from wood, and the chemi-
cal recovery cycle that recovers cooking chemicals and
produces process heat. Heat recovery is applied to
reduce energy consumption, and in a pulp mill this is
achieved mainly in a secondary heat system (HWWS)
consisting of heat exchangers and tanks. The HWWS
has four main functions:
1. To supply warm and hot water to the process, for
example to dilute the pulp.
Qxs
Q (kW)
∆Tmin
Excess heat of
low temperature
Excess heat of
high temperature
T
(°
C
)
b
ds, which could be used for Qxs, varies. (b) Shifting the tank curve left
2. To heat cold process streams in heat exchangers.
3. To cool hot process streams, thus producing warm
and hot water.
4. To act as a heat buffer when the heat demands and
supplies vary.
The heating and cooling demand in the HWWS in a
pulp mill is illustrated in Fig. 1a. Arrows illustrate heat
recovered by heat exchange. The remaining heat varies
with temperature and is presently cooled away. When
the hot streams are cooled, more warm and hot
water than needed is produced; this overproduction
can be seen as cold utility. It is, however, not
necessary to use the HWWS as a cold utility. Instead
the cooling could be separated from process water
heating.
Shifting the cold CC leftwards would increase Qxs at
high temperatures, but also decrease the DTmin and
increase the area required for heat exchange, Fig. 1b.
below.
3. Identify the process demands and the cooling
demands from the cold streams.
4. Construct the cold CC from the cold streams for the
process demands.
5. Calculate the theoretical target for the maximum
excess heat at a given DTmin.
6. Replace the cold CC with the ‘‘tank curve’’. The tank
curve is defined below.
7. Vary the number of tanks and their temperature
levels to maximize the amount of excess heat.
Step 1. Construct the hot CC from all hot streams
in the HWWS.
Steps 2–4. In our method the cold CC represents only
net process demands of hot or warm fresh
water. All cold streams� start temperature
is set to the inlet temperature of the fresh
water. The target temperature is set to the
process demand temperature. Defining the
Q (kW
rom th
rs.
l cold
R. Nordman, T. Berntsson / Applied Thermal Engineering 26 (2006) 363–373 365
1. Identify the HWWS hot streams and construct the
hot CC.
2. Identify the HWWS cold streams.
T
(°
C
)
0
10
20
30
40
50
60
70
80
90
0 5000 10000 15000
The tank curve for one tank level (at 85 ºC)
The original cold curve f
demands of the final use
Fig. 2. Tank curves are constructed based on knowledge of the origina
Note that the DTmin decreases, although QC is the same.
The purpose of this shift would be to make the excess
heat at high temperature usable.
4. New design method
The method developed is based on composite curves
[7]. In addition to the hot and the cold CC�s, a third
curve is introduced, called the ‘‘tank curve’’. The steps
in the method are listed and explained in more detail
and their temperature-wise placement.
)
20000 25000 30000 35000
TTL 1, curve 1
Two tank curve with two tank levels
e
composite curve, and look different depending on the number of tanks
cold CC in this way the slope will contin-
uously increase, since the slope equals
1/F * Cp (Fig. 1a).
Step 5. Shift the cold CC leftwards until a user set
DTmin is reached. The part of the hot CC
that overshoots on the right of the cold
CC is the theoretical maximum of Qxs
(Fig. 1b).
Step 6. Heating of fresh water to the different TTL�s
in the system composes the tank curve. All
fresh water is heated from the fresh water
inlet temperature to the temperature levels
where tanks are placed in the system. The
number of stream intervals in the tank curve
is therefore equal to the number of tanks in
the HWWS. The tank curve has the same
T = 85 ºC
TTL 1, curve 2
TTL 2, curve
1 and 2
trarily on the cold CC. The amount of excess heat that
can be made available differs depending on the combina-
6. Case study
6.1. General description
To demonstrate the new method a case study has
been carried out in the Storaenso Skoghall mill. It is
an integrated board mill with both CTMP and kraft
pulp, and 550,000 tonnes board/yr capacity. The pro-
duction of CTMP is 200,000 tonnes/yr and the kraft
pulp production is 290,000 tonnes/yr. This study has
only covered the kraft pulp plant. The Skoghall mill is
considered a low-water-usage mill.
The opportunities for a grass-root and/or retrofit
366 R. Nordman, T. Berntsson / Applied Ther
tion of TTL�s (Fig. 2). For each combination of temper-
atures, a tank curve will replace the original cold curve.
This curve is then shifted leftwards in the diagram until
DTmin is reached. By systematically checking the value
of Qxs for all TTL combinations in the HWWS, the
maximum Qxs is reached for a specified number of
tanks.
A theoretical optimum of Qxs is not necessarily an
economic optimum; thus an economic evaluation must
be done for different DTmin�s. A small DTmin will give
more excess heat but the heat exchanger area needed
to release this heat will increase. Hence there is a
trade-off between investment costs for new heat
exchanger area, and expected savings from the released
excess heat. It could also be practical to place a tank
at a temperature level where there is a process demand
even though it is not optimal to maximize Qxs. In this
way, mixing of the tank water with cold water to reach
the demanded temperature is avoided. More details on
the economic evaluation are presented in the case study.
The method has been implemented in a Visual Basic
code. Computation effort for a four-tank problem
involves approximately 3.2 · 107 function evaluations,
requiring about 780 CPU seconds on a 2.4 MHz
Pentium 4 PC.
5. Applications using excess heat
Two ways to use the released energy were investigated:
to replace live steam in a non-conventional evaporation
plant [1], or to provide district heating. A third, but not
studied, way to use the Qxs would be heat pumping.
Raising steam from the excess heat could replace live
steam in the evaporation plant. Because of the low tem-
perature of the steam raised, it will not be possible to use
it in all evaporator effects. The total heat needed for
evaporation, Qtot, can thus be expressed as (see Fig. 3
for designations):
heat demand as the original cold CC. The
two curves coincide at the TTL�s, as well
as at the start point.
Step 7. Maximize the amount of excess heat by
varying the number of tanks and their
temperature levels, as well as the DTmin.
One tank level always has to be placed at the highest
temperature to supply the warmest demands. If only one
tank were installed in the system, all water would have
to be heated to the highest temperature. With two
TTL�s, one tank still has to be placed at the highest
temperature, while the second TTL can be placed arbi-
Qtot ¼ QI � N I ð1Þ
or
Qtot ¼ QII � N II þ Qxs � n ð2Þ
The reduction in live steam demand, DQ, by using Qxs
is then
DQ ¼ QI � QII ¼
Qtot
N I
� Qtot � Qxs � n
N II
ð3Þ
For the case when NI = NII = N:
DQ ¼ Qxs � n
N
ð4Þ
Here we introduce the factor C as the ratio of saved live
steam (DQ) to the amount of Qxs when live steam is re-
placed in non-conventional evaporation. The factor C
typically takes values between 0.6 and 0.9 but values
greater than unity can occur, e.g. when the surface con-
denser operating temperature is lowered. Excess heat is
then allowed to cascade through more evaporator units.
The reduction of live steam results in lowered fuel con-
sumption and lowered electricity production, since less
steam runs through the steam extraction back-pressure
steam turbine.
Using the Qxs for district heating requires that the
plant is located close to a community that already has
a DH network; otherwise the investment cost would
be unrealistically large for almost any value of Qxs.
QXS
QIIQI
NI
n
NII
Fig. 3. Scheme over traditional (left) and excess-heat-driven (right)
(also known as non-conventional evaporation) evaporation.
mal Engineering 26 (2006) 363–373
design of the HWWS were investigated, using the new
T
1
1
1
1
1
7 75 1500
Thermal Engineering 26 (2006) 363–373 367
method, with the aim of releasing Qxs at temperatures as
high as possible. Economic consequences for different
use of the released Qxs were studied. The existing
HWWS contains four tank levels with water at 45, 65,
75 and 85 �C. These tanks supply the warm and hot
water demanded in the process. The system also handles
the cooling demands in the process, about 41 MW. The
overproduction of hot water is drained away. Stream
data for the HWWS are taken from previous work by
Table 1
Stream data representing the HWWS streams in the case study
Stream name Type
Clean condensate from condenser Hot
Whitewater cooler after ClO2 reactor Hot
Stripper condensate cooler Hot
Wash liquor cooler, wash filter Hot
Scrubber Hot
HX D007 Hot
Evaporator 1 Hot
Evaporator 2 Hot
Evaporator 3 Hot
Pre-evaporator Hot
Condensing of relief vapors Hot
After HX D004, steam Hot
After HX D004, supercooling condensate Hot
Wash press after ClO2 Cold
Wash press after Q bleaching Cold
Oxygen bleaching Cold
Mixing plant Cold
Internal and external heating Cold
Tank B-900 Cold
WW 65 unbleached pulp distribution Cold
WW 45 unbleached pulp distribution Cold
R. Nordman, T. Berntsson / Applied
Bengtsson [8] where the streams were defined. Table 1
lists these streams and the corresponding cold and hot
CC�s are shown in Fig. 1a. Because only process streams
that have a heating demand are included, the CC�s are
not balanced. Hot streams that are not fully utilized in
internal heat exchange to reach its target are cooled to
the target temperature.
6.2. Excess heat potential
A total of 16 combinations of DTmin and tanks
(DTmin of 1, 3, 5 and 10 K and 1, 2, 3 or 4 tanks) were
optimised to maximize Qxs. The results showed that al-
ready with two tanks 100% of the theoretical maximum
Qxs was reached. More tanks would therefore not bene-
fit excess heat. The minimum temperature of Qxs is di-
rectly related to the amount of excess heat available;
when Qxs increases, the lowest temperature of the excess
heat also decreases.
6.3. Sensitivity analysis
The sensitivity of choosing the tank temperatures in
an HWWS with two tanks at different temperatures has
been analysed. One TTL is chosen at the highest tem-
perature demanded in the system, which for the case
study is 85 �C. The second TTL can be chosen arbi-
trarily on the cold CC. Fig. 4 shows the amount of
Qxs as a function of the second tank temperature level
for different DTmin�s. When DTmin is large, no excess
heat can be released, which is indicated by zero value
in the diagram (unpinched problem). In a system with
a small DTmin the choice of temperature level need
7 65 8570
7 85 5000
7 55 4210
7 65 2150
7 45 1410
start (�C) T target (�C) Q (kW)
70 20 3060
66 57 810
10 48 5390
02 89 2860
67 60 5830
10 100 1600
76 42 6340
68 67 24,530
67 48 830
77 55 18,960
80 50 3270
23 122 980
22 43 1960
7 85 6860
7 85 6860
not be very exact, because of the flat optimum. With
larger DTmin it becomes increasingly important to
choose the right temperature level. The optimal TTL
for achieving a high Qxs decreases as the DTmin in-
creases, and the optimal TTL is found where the system
pinches.
6.4. Economic evaluation
Good industrial practice means that DTmin values
lower than 5 K would not be used [9]. Therefore, in
the economic evaluation, only DTmin�s of 5 K and
10 K were evaluated, although even larger amounts of
Qxs could theoretically be released by smaller DTmin.
The economic evaluation for the different design alterna-
tives is based on the net annual savings method:
Net annual savings (NAS) = Income from sold excess
heat � Income losses because of lost electricity produc-
tion � Annualized investment costs.
Positive values of NAS mean that the investment is
recommended.
The annualised investment costs are expressed as
I � a ð5Þ
45
mper
perat
Elect
Thermal Engineering 26 (2006) 363–373
0
1000
5 15 25 35
Te
Fig. 4. Usable Qxs for different DTmin as a function of the tem
Table 2
Cost parameters for the scenarios used in the economic evaluation
Scenario Oil price (US$/MWh) Bark price (US$/MW h)
2000
3000
4000
5000
6000
7000
8000
Q x
s
(kW
)
∆Tmin = 15K
∆Tmin = 10K
∆Tmin = 5K
∆Tmin = 3K
∆Tmin = 1K
368 R. Nordman, T. Berntsson / Applied
where I is the capital investment, a is the annuity factor.
The investment is annualised using an annuity factor
based on interest rate, i, and assumed economic lifetime,
n. The annuity is expressed as [14]:
a ¼ i
1� ð1þ iÞ�n ð6Þ
Annuity factors of 0.1 and 0.25 were used in this case
study. Four different energy market scenarios [10] were
used to evaluate the economic consequences. The eco-
nomic parameters for each scenario are listed in Table 2.
6.5. Investment costs
A ‘‘grass-root’’ design would be an alternative only if
one has to invest in a new HWWS. In the economic eval-
uation of the grass-root design, only the extra cost for
building a system that releases the excess heat has been
included. In the retrofit cases the total investment cost is
included. The calculations include cost for new heat
exchangers and new tanks, and the extra cost of building
non-conventional evaporation compared to traditional
evaporator design. Piping cost is calculated as 40% of
FOB price of new heat exchangers [11]. The cost for
the steam reformer is in each case calculated as a heat
exchanger, Table 3. The volume of the tanks is chosen
1 26.9 13.4 38.7
2 22.0 13.9 40.3
3 29.3 18.2 52.5
4 41.1 26.4 76.2
55 65 75 85 95
ature (˚C)
ure of the second tank level in a system with two tank levels.
ricity retail price (US$/MW h) District heat price (US$/MW h)
to carry 24 h of hot water demand. A new evaporation
plant is so costly that investment is only justified if
new evaporators have to be built anyhow. Instead the
marginal cost of building a non-conventional evapora-
tion is calculated. As reference, a traditional evapora-
tion plant with six effects and a surface condenser at
60 �C has been used. Other ‘‘traditional’’ configurations
exist, e.g. seven effects with surface condenser at 40 �C,
but the one chosen represents the most common design.
The non-conventional evaporator plant is suggested to
have seven effects with excess heat used in the last four
or five effects, depending on the temperature of the
excess heat.
Cost equations used are:
Heat exchanger costs [12]:
Costð$Þ ¼ 30800þ 1644 � A0.81HX ð7Þ
Tank costs [13]:
Costð£Þ ¼ 2.5 � 1200 � V 0.55Tank ð8Þ
Evaporator costs [1]:
Costð$Þ ¼ 8.5� 105 � nþ 333 � AEvap ð9Þ
Cost of lost electricity production:
Costð$=yrÞ ¼ Elost � top � Cel ð10Þ
20.2
20.8
26.0
35.8
R. Nordman, T. Berntsson / Applied Therm
Table 3
Investment cos