Sei sulla pagina 1di 3

System analysis of the refrigeration system at Philip Morris Manufacturing Berlin

to develop cost and energy saving potentials


Franziska Dan September 2018

Abstract
With a share of 41% of the total electricity demand of the technical building equipment excluding the air-conditioning
and ventilation systems of the production location Berlin, Philip Morris Manufacturing’s historically grown refriger-
ation structure represents a decisive starting point for the identification of energetic optimization potentials. Within
the framework of this study, holistic solutions for the two main refrigeration plants will be developed considering
all cost and energy intensive main components. As a tool for analyzing the complex interaction of the different
system components, a dynamic annual simulation is used. By means of this simulation, various control strategies
for optimizing system temperatures and switching specifications for the available recooling systems and chillers are
developed. In addition to diverse control concepts, which can be assumed in any case as economically viable, also
the potential of various reconstruction measures with a comparatively higher capital expenditure will be assessed as
another emphasis of investigation. If all the recommended measures derived from this are implemented, significant
energy and cost savings of up to 19% and 22%, respectively, can be achieved within the existing refrigeration supply.
Keywords: control concepts, recooling systems, water treatment costs, turbo compressors, cost and energy efficiency

1 Introduction
A total of seven refrigeration plants cover the year-round The simulative and analytical procedure is explained
high refrigeration demand at PMMG Berlin. The two re- in part and the main results of the respective partial
frigeration centers KA-001 and KA-008 take over 80% of analyzes are presented.
this total demand, which was 17.56 GWh in the refer-
ence year 2017. Because they have a higher system ef- 3.1 Control-based optimization
ficiency compared to the other centers, they are deliber-
ately prioritized and can therefore be considered as the Refrigeration machine’s part load behavior
two main refrigeration centers. The ability to combine Based on a detailed analysis of the individual plant char-
dry and wet recooling systems as well as the possibility acteristics of the three identical refrigeration machines
of variable setpoint specification gives the basic prerequi- (RM) with turbocompressors available in KA-001, the pos-
sites for exploiting further optimization potentials. itive influence on efficiency (EER) by
 favorable partial load behavior,
The main objective of this study is to identify further
savings potentials for the two main refrigeration plants  lower cooling water inlet temperatures tin and
and to derive specific recommendations for improved op-  higher cold water flow temperatures tf low
erating practices. The actual problem arises from the
complex interaction between the components. Because a could be identified and quantified. The different parame-
single component-wise optimization is not automatically ters influencing the efficiency of the refrigeration machine
equivalent to an actual system optimization due to reper- can be seen from the EER matrix highlighted in color in
cussions. Rather, potential improvement measures of one Table 1. The best possible EER value of 11.12 results from
component can simultaneously cause negative energetic tin =16◦ C and a load of 45-50 %, while under unfavorable
and / or monetary influences elsewhere. The development operating conditions at tin =24◦ C and under full load only
of holistic solutions to the system-related cost and energy an EER value of 5.95 can be achieved.
saving is therefore the overarching task of this work. Table 1: Matrix for describing the refrigeration machine efficiency
(EER) for tf low ≈ 6◦ C generated by data-based function lines.
(orange: lower - green: higher efficiency)
2 Method
The entire system can be visualized as a gear with a va-
riety of screws, in which the influence of reciprocal ef-
fects must always be considered. Oriented on this, the
methodology must also move to an interplay of analyti-
cal, numerical and practical approaches.
Numerical problem solving is done using the simulation
software Modelica/Dymola. The first design part com-
prises the modeling of the individual components and
their interlinking to an overall model. The correspon-
dence between model and reality was ensured by subse-
The comparison of these two high-contrast operating
quent validations as a basic prerequisite for the simulation
states with an efficiency difference of 87% illustrates the
in the next work step. It represents the tool for dynamic
great potential for optimization through intelligent con-
and not only static comparative analysis between the ini-
trolling of the mentioned influencing variables. In view
tial state and the operating variations to be examined.
of the high RM efficiency at the lower capacity range, it
Only by this numerical approach, the reciprocal influence
may be advisable in certain cases to switch on another
of all relevant system components by the dynamic speci-
refrigeration machine even before reaching the maximum
fication of ambient and load conditions can be examined
capacity range of the operating one. However, to maintain
and thus, an actual system optimization is ensured.
the highest possible system efficiency, the entire system,
In addition, analytical and practical approaches are also including the energy-intensive pumps must be considered.
essential. Based on this combined consideration method- The enormously significant pump energy of 51kW per RM-
ology, a wide range of different control strategies and con- cycle has a massive impact on the position of the switch-
version recommendations has been developed. ing point optimum. Using a simulation-based systematol-
ogy, the optimal switching points of the second and third
3 Optimization measures chillers could be differentiated according to tin und tf low .
In Figure 1 the developed systematology is illustrated us-
This section provides an overview of the recommended op- ing the example of the ideal switching ranges between the
timization measures developed for the two main centers. second and third refrigeration machine.
In the left diagram of Figure 1: Optimal switching points depending on tin snd tf low using the example of KM2x to KM3x
Figure 1, the ∆Ptotal,3−2 -
curves of three different
cooling water inlet tem-
peratures tin =18/22/26◦ C
at tf low =6◦ C are applied.
They are determined by
driving the full bandwidth
of the RM2→3 power spec-
trum between 1040kW and
2320kW in the modeled and
simulated overall system.
The difference ∆Ptotal,3−2
is given by subtracting the
total power of two re-
frigeration machines (RM2x )
from the total power of Hier kann ein Bild hin
Figure 2: Left: Procedure for determining the optimal twet−recooler default values using the example
of the cluster area tW etBulb,10...12◦ C at Q̇0,1000kW and tf low,6◦ C .
Right: Extract of three optimized twet−recooler default curves for tf low,6◦ C .
three chillers (RM3x ) un-
der the same conditions.
Due to the intersection
with ∆Ptotal,3−2 =0kW as
an indicator for optimized
switching, it can be seen
that e.g. at tin =26◦ C
it makes sense to con-
nect a third chiller from
Q̇0 =1926kW. This could re-
sult in Ptotal savings of
37kW in the maximum case
at Q̇0,max,2x =2320kW.

From the right diagram it is recognizable that also un- mined. The principle of clustering usedP12for this purpose
der other tin - and tf low -conditions, the connection of the differentiates between the Q̇0 -ranges i=1 i · 250kW, the
third chiller before reaching Q̇0,max,2x is energetically rec- P13
tW etBulb -ranges i=0 i + 10◦ C and tf low . For each of the
ommendable. The greater potential tends to be at higher 156 clusters per tf low -variant, the default wet recooler
tf low - and rising tin −temperatures. outlet temperature runs the range from 16◦ C to 27◦ C.
Through the associated Ptotal -curve, an individual opti-
Dry recooling operation mum can be assigned to each cluster. Figure 2 illustrates
The focus of the second of altogether six different con- the procedure using the example of one cluster area. The
trol strategies in total is on the development of intelligent three determined optimized default values from the left
switching criteria between dry and wet recooling. With diagram become transferred to the right diagram along
regard to the often high water treatment costs in con- with other optimization results. The extract illustrates
nection with the wet recooling operation, a considerable the tendency that with increasing wet bulb temperature
energy and particularly monetary saving potential can be and cooling load, the optimum wet recooler outlet tem-
exploited by a greater utilization of the capacity limit of perature for the entire system should also be increased.
dry recooling plants. However, the amount of savings de-
pends decisively on the specific plant characteristics and Flow temperature increase
the individual cost factors and requires a differentiated
case-by-case analysis depending on the type of plant. For In contrast to the turbo compressor technology in KA-001
the refrigeration plant KA-001 it could be shown that by chillers, the screw compressors in KA-008 offer less vari-
optimized switching criteria and the associated improved ation possibilities. In the partial load case, no signifi-
control concept an extension of the annual operating time cant increase in efficiency is detectable and with regard
of the dry recoolers by at least 1400 operating hours can to the cooling water temperatures, a deviation from the
be achieved. The fundamental approach of this concept is manufacturer’s default values should be avoided. The de-
based on harmonizing both a high dry recooler operating velopment of a concept for cold water flow temperature
time and high EER-Values through a sliding tdry−recooler increase represents a reasonable way to increase efficiency.
outlet temperature. For a characteristic period of just The flow temperatures have been permanently set at 6◦ C
under 1000 hours, an increase in total plant efficiency of on the grounds that in principle, three of the numerous air
18.1% and a corresponding reduction in operating costs conditioning and ventilation systems supplied by KA-008
of 22.4% was demonstrated by means of the simulation. require dehumidification. However, a detailed analysis of
the critical air conditioning systems has shown that only
Wet recooling operation in 18% of the almost 7900 operating hours a dehumidi-
From the previous analytical and simulative studies it is fication process would have been required. In all other
known that the chiller’s efficiency can be positively influ- times, the flow temperature 9◦ C would suffice to temper
enced by lower cooling water inlet temperatures. Against the supply air and even a certain degree of dehumidifica-
this background it can be stated that the previously stat- tion could be achieved with tf low >6◦ C. Along with such a
ically at 24◦ C or 26◦ C set twet−recooler -values in the wet flow temperature increase, according to the manufacturer,
recooling operation require the development of a dynamic a promising EER improvement in the order of ∆EER=0.5
control strategy. In order to meet the objective of increas- could be achieved.
ing system efficiency, a large number of influencing factors
must be taken into account when choosing the optimum By linking the relevant cooling valve positions as a funda-
twet−recooler default value. The problem solving therefore mental approach of the developed concept, it is ensured
takes place with a simulative solution approach. that only in case of acute dehumidification, a gradual re-
duction from the new default value tf low =9◦ C is allowed.
Through a systematic twet−recooler -variation in the mod- The control concept has proved suitable during operation.
eled refrigeration system, the optimal control default val- Based on a dynamic annual simulation, an increase of the
ues for every conceivable operating condition were deter- overall system efficiency of KA-008 by 3% was proven.
Further control-based optimization
In addition, two further control concepts were developed than three times the required capacity of a single
within the framework of the present study. On the one consumer group to be supplied exclusively on weekends.
hand, it could be shown that the free cooling operating This problem could be circumvented by supplying the
time can be extended by 11% due to optimized swiching critical consumer group with a local cooling system
criteria between free cooling and dry recooling operation. instead of using the KA-008 cooling network as before. A
The simulative study of a characteristic simulation sam- corresponding possibility is offered by the installation of
ple period of 550 operating hours shows that the improved an efficient climate cabinet. Implementation has already
control concept can achieve an efficiency increase of 8% taken place and the energy and monetary benefits
for this example period. for KA-008 have been quantified by means of another
simulation. Based on the simulation results, the weekend
On the other hand, the optimization potential regard- shutdown results over the year in a 3% increase in
ing the existing outlet temperature control for the only efficiency and a 13% reduction in operation costs.
stepwise controllable wet recooling plant in KA-008 was
analyzed. The present recooling plant can be operated in Refrigeration machine exchange
wet as well as dry mode by means of five different stages. The model developed for the purpose to describe the spe-
Each stage is assigned an individual combination of fan cific characteristics of such turbocompression refrigera-
number, fan stage and spraying. A simulation was used to tion machines represents the basis for another emphasis of
investigate the influence of a different stage sequence. As investigation within the main refrigeration plant KA-008.
a result, the superiority of a new stage constellation could System efficiency could be significantly increased if the
be proven. By adding an energetically and economically outdated existing refrigeration machines were replaced by
reasonable intermediate stage with a lower spraying pro- the more efficient and rarely operated turbo compression
portion, KA-008 can achieve an annual efficiency increase refrigeration machines of another plant. The results of
of 4% and an operating cost reduction of 7%. the completed annual simulation show that the achiev-
able increase in efficiency within KA-08 amounts to 60%.
3.2 Modification measures Compared to all other investigation measures, the saving
effect of this reconstruction measure is greatest.
In addition to the economically viable control concepts,
the potential of various reconstruction measures with a
comparatively higher capital expenditure was assessed. 4 Conclusions from the dynamic year simulation
SGV - volume flow controlled refrigeration machine As a final result of this study, the overall energetic and
The focus of this partial analysis is the so-called SGV- monetary effect should be demonstrated by implementing
retrofitting of the turbo-compression refrigeration ma- all the recommendation measures summarized in Table 2.
chines in the main refrigeration plant KA-001. In the Table 2: Overview of all recommended implementation measures. ...
course of such SGV-modification, the power control is im- (Control measures in the upper part - rebuilding measures ...
plemented by means of a previously not executable vari- in the lower part of the diagram.)

able volume flow control for the purpose of saving pump KA-001 KA-008
energy. Simultaneously, a deterioration of the hitherto X RM part load behavior X drywet recooling
high chillers efficiency in part load cases must also be X drywet recooling X tf low increase
taken into account. For a precise quantification of this X tdry−recooler -optimization X stage constellation
non-negligible influence, a systematology was developed X twet−recooler -optimization
X free coolingdry recooling
to mathematically describe the modified thermodynamic
X SGV X weekend shutdown
behavior. The simulation results for this adapted refriger-
X RM exchange
ation system reveal the overall positive impact on the to-
tal efficiency. An increase of 5.3% can already be achieved
by modifying one refrigeration machine circuit thanks to The achievable savings in the main refrigeration plants
the significantly reduced power consumption for the mod- examined are placed in the context of the total electrical
ified pump and despite the simultaneous loss of efficiency consumption of all seven refrigeration plants measured
on the part of the chiller. in the reference year 2017. The associated operating
costs consist in addition to the electricity costs also of
Weekend shutdown the calculated water treatment costs. Based on these
In the other main refrigeration plant KA-008, the savings consumption and cost values, the classification of the
potential by adjusting the distribution of demand were developed savings potential related to the entire cooling
investigated. Due to minimum loads and distribution supply at PMMG in the initial state takes place. The
losses, the cooling capacity actually provided was more final results are presented in Figure 3.

The totality of all


Figure 3: Energetic and monetary overall effect of all recommendation measures recommendations devel-
oped allows significant
reductions in electrical
energy consumption
from 5.7 to 4.6 GWh
and operating costs
from 896 to 697 TA C.
Based on the results of
the annual simulations,
annual energy and mon-
Hier kann ein Bild hin etary savings of 19%
and 22%, respectively,
can be demonstrated. In
each case, an economic
viability of the control
measures and a fast
amortization period of
a maximum of 3.2 years
can be expected for the
conversion measures.

Potrebbero piacerti anche