- Email: [email protected]
Decentralised cooling in district heating network: Monitoring results and calibration of simulation model
Energy and Buildings 43 (2011) 3311–3321
Contents lists available at ScienceDirect
Energy and Buildings journal homepage: www.elsevier.com/locate/enbuild
Decentralised cooling in district heating network: Monitoring results and calibration of simulation model Seksan Udomsri a,∗ , Chris Bales b , Andrew R. Martin a , Viktoria Martin a a b
Division of Heat and Power Technology, Department of Energy Technology, Royal Institute of Technology (KTH), 100 44 Stockholm, Sweden Solar Energy Research Center (SERC), Högskolan Dalarna, 781 88 Borlänge, Sweden
a r t i c l e
i n f o
Article history: Received 17 December 2010 Accepted 1 August 2011 Keywords: Thermally driven chillers Decentralised cooling District heating network Monitoring results Calibration of simulation model TRNSYS
a b s t r a c t This article presents the monitoring results of a thermally driven chiller (TDC) driven by district heat from a network supplied by a centralised combined heat and power (CHP) fired with municipal waste. The main objective of this article is to analyse the monitoring results obtained from the demonstration and calibrate a system model that is later used for parametric studies in order to find improved system design and control. The calibration of the system model was made in three stages and all the energy performance figures were within 4% of the measured values. Results show that the TDC system is capable of providing maximum thermal and electrical COP’s of 0.50 and 4.6 respectively during the hottest period. For the complete monitoring period during the summer of 2008, the figures were 0.41 and 2.1. The lower figures were due to continuous pump operation inside the TDC even during periods of no cold production and a period when no cold was produced. However the internal pumps inside the TDC have been removed in the new version TDC to increase the electrical COP. System simulation and parametric studies will be employed to further determine how the electrical COP can be improved. © 2011 Elsevier B.V. All rights reserved.
1. Introduction 1.1. Background and objectives The demand of comfort cooling has dramatically increased in the EU over the past few years, even in cold climate like Sweden, and district cooling market is expected to expand tremendously in the next decade and beyond. In Sweden, the demand for cooling has increased over the past few years despite its upper northern latitude [1,2]. However, the largest cooling demand has always been found in summer period where low or no heating load exists. Efforts are being made to enhance thermal power plants via absorption technologies in order to increase the exploitation of heat and utilize low-grade heat for cooling production. Various alternative sources of cooling production have been explored intensively to replace mechanical chillers towards reducing electrical energy consumption and CO2 emissions [1–4]. One concrete example of particular significance is decentralised cooling in district heating networks. It is an attractive approach where the cooling is provided on the demand side and consumers can deploy the system themselves to produce cooling by using heat from district heating network.
∗ Corresponding author. Tel.: +46 8 790 6140; fax: +46 8 20 41 61. E-mail addresses: [email protected], [email protected] (S. Udomsri). 0378-7788/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.enbuild.2011.08.001
This article presents the monitoring results and calibration simulation of the demonstration system of decentralised cooling in district heating network, which is one of the 11 demonstration systems installed and monitored within the EU-PolySMART project (www.polysmart.org) [5]. The system employs a ClimateWell [6] thermally driven chiller (TDC) supplied by heat from the district heating system. The PolySMART project aims were to develop a set of technical solutions for a new market segment of polygeneration, in particular the market for small tri-generation systems (combined production of electric power, heat and cooling). The key components of these systems always include the combined heat and power (CHP) plant along with the TDC. There are different approaches of composing the CHP with TDC [5]: (i) centralised CHP and centralised TDC in combination with a heating and cooling network, (ii) centralised CHP and decentralised TDC in combination with a heating network, and (iii) decentralised CHP and decentralised TDC both on the demand side. Each approach can be applied under different conditions depending upon the source of generation, consumption, and applications. However, this article reports on a combination of centralised CHP and decentralised TDC in district heating network. District heating can be described as rational and environmentally friendly method to heat residential and commercial buildings, etc. District heating is very common heating method and available throughout Sweden. The CHP unit generates electricity and heat where the heat can be distributed in term of steam and district heating in the district heating network.
3312
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Nomenclature CHP COP Cp DH E Eff HEP LHV ˙ m P PE Q RH SP T TCA TDC Tim V Dpi
combined heating and power coefficient of performance specific heat capacity district heating electricity consumption efficiency high efficiency pump lower heating value of a fuel mass flow power primary energy thermal energy flow relative humidity subproject temperature Thermo-Chemical Accumulator thermally driven chiller time volume flow pressure drop in each circuit
Subscript Air air BOP balance of plant chilling circuit of TDC Cc Cdn cold distribution system for end-use N Dc driving circuit of TDC el electricity feed line (considering as the one leaving the heat Fl source, and thus the hottest line) high efficiency pump HEP Hr heat rejection system Meas measurement OA outside ambient Pump pump recooling circuit of TDC Rc RH relative humidity Rl return line (considering as the one returning to the heat source, and thus, the coldest line) Sim simulation Sys system Tdc thermal driven cooling thermal energy Th ti time
As often found in Europe, especially Sweden, cogeneration facilities are an important contributor to the overall energy mix, in particular for heat delivery in wintertime. One important feedstock for the CHP unit is municipal solid waste (MSW). However the capacity of district heating is not fully utilized during non-heating period or the summer time. This means the CHP plant operates relatively low capacity and low efficiency at consequently. Thus it is important to find new applications for the heat produced during this period e.g. using heat as energy carrier for distributed smallscale heat driven machines. The commercial TDC usually employs heat from steam, hot water or from a solar thermal system or even directly from combustion gases as a heat source. The main application is comfort cooling or for industrial processes. There are several technologies of absorption chillers commercially available today, e.g. a standard absorption chiller using LiBr/water or NH3 /water and salt-water absorption chiller and/or chemical heat pump. Absorption chillers are more common at medium or larger
scale, while small scale TDC are in process of becoming commercial. Chemical heat pump is new and promising technology in which it is capable of powering its system with low temperature heat sources. Salt-water solutions such as lithium chloride (LiCl)/water, sodium sulphite (Na2 S)/water, and calcium chlorides (CaCl2 )/water, etc. have been used [7–10]. The TDC in this study is a recently commercialised chemical heat pump using LiCl/water as a working fluid pair [8,11]. The objective of the demonstration was to demonstrate the use of the ClimateWell chiller in distributed cooling with centralised CHP in order to develop the best system configuration for the TDC using a particular form of chemical heat pump, the ThermoChemical Accumulator (TCA). The main objective of this article is to present the results obtained from the demonstration as well as to calibrate a system model that was later used for parametric studies in order to find improved system design and control. Results obtained from monitored data will be presented and verified against simulated results obtained via dynamic modelling with TRNSYS (TRaNsient SYstem Simulation program). 1.2. Thermally driven chiller (TDC) A chemical heat pump or Thermo-Chemical Accumulator (TCA), patented in 2000 [8], has been employed and installed in this project as a TDC unit. It has been developed and is sold by a Swedish company ClimateWell AB. It is a three-phase absorption chillers/heat pump that is capable of storing energy internally with high energy density in the form of crystallized salt (LiCl) with water as refrigerant. In principal, the process operates under vacuum condition like in standard absorption chillers using LiBrwater. The TCA is however significantly different from traditional absorption chillers in that it works in batch mode with relatively long cycle times (>6 h). The common heat sources for this are district heating, waste heat from cogeneration and solar thermal collectors. Fig. 1 shows diagram of the ClimateWell chiller including major components that is referred to as the TDC system in the rest of this article. The triple-state process, so called because it uses solid, solution and vapour at the same time, makes it particularly different from other chemical heat pumps or standard absorption processes (two phase) [11]. It consists of two identical units, so called barrels, that work together to provide quasi-continuous operation (see Fig. 1). Each barrel consists of a reactor and a condenser/evaporator that is connected by the gas pipe or vapour channel. The reactor, called generator/absorber in normal absorption chillers, contains sorbent/hygroscopic salt solution (LiCl), while the condenser/evaporator contains pure water (refrigerant). The system operates intermittently with a charge phase followed by discharge phase. In the charging phase, salt-water solution is heated by a thermal source via a heat exchanger in the reactor and the solution becomes steadily more concentrated. This can be continued until solid is formed. The solid (crystals) are physically restrained from being transported from the reactor and causing clogging. At the same time water is evaporated and steam is released to the condenser/evaporator. In discharge process, a reversed process takes place, with a heat exchanger transferring heat from the building to the water, which evaporates and is transported to the reactor. The water gets absorbed by the concentrated salt, either in solid or solution form depending on the state of charge, and heat is released and transferred to an external circuit via a heat exchanger [6,11,12]. The technology has been developed through five different generations, each with its own particularities that make it significantly different in terms of operation from its predecessors. In this study, two versions of the TDC were analysed: the 4th generation as installed in the demonstration system, subproject 1b (SP1b) in Sweden, and a 5th generation as was installed in the Madrid, Spain
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
3313
Fig. 1. Working principal of the TCA unit with major components included [left] and drawing of ClimateWell solar chiller with two barrels [right] [6].
(SP1a). The 5th generation is the one currently sold (2010) is implemented in system simulations and parametric studies of SP1b that are described in Decentralised cooling in district heating network: system simulation and parametric study [13].
2.2. General boundary conditions for the plant
1.3. Methodology
Table 2 shows general boundary conditions and components of the plant, mostly for CHP, TDC and heat rejection unit.
The demonstration system of decentralised cooling driven by district heating network was installed, monitored and calibrated within the course of the project. The calibration simulation model was made and compared again monitored data. Performance figures for evaluation of the results were defined and criteria was set that the model should predict the use and delivered energy quantities of within 4% of measured values. The following steps of system calibration and simulation is presented in Table 1.
evaluated system is thus only a part of the total cooling system and does not operate during the winter.
2.3. Definitions of COP General formula of thermal and electrical COP’s is defined in this section. The thermal COP for both system and TDC unit can be Table 1 Steps of calibration in the system study and simulation. Step
Calibration of system model against monitored data
1
The calibration of the system model was made in three stages: (i) estimation of parameters based on manufacturer data and dimensions of the system; (ii) calibration of each circuit (pipes and heat exchangers) separately using steady state data points; (iii) and finally calibration of the complete model in terms of thermal and electrical energy as well as running times, for a five day time series of data with 1 min average data values. Subsystem calibration. The three subsystems for the driving, recooling and cooling circuits to the TDC were calibrated against the measured data in terms of energy balance using a range of steady state values from different operating states. The UA-values of the heat exchangers and of the pipes are the parameters that were varied in the calibration process. Determination of the power use of the fan in the dry cooler as a function of the air flow. This was carried out based on the identified dry cooler parameters, measured weather data and inlet/outlet temperatures together with average electrical power for quasi steady state periods. The flow temperature from the cooling distribution loop coming from the air handling units (TCdnFl ) was analysed and a simple correlation between TCdnFl and ambient temperature was derived. The complete system model was then calibrated against a five day dynamic measurement sequence from a hot period. As the TDC has significant internal thermal storage, the starting state of charge had to be determined for each of the two internal storage units (barrels). The main criteria for calibration were the thermal and electrical energies of the whole system. The parameters that were varied in order to gain a good fit were the control parameters for TDC, electrical power of components, UA-values for the TDC heat exchangers and losses from the internal stores.
2. Demonstration plant and performance figures 2.1. Description of the demonstration system The demonstration plant, subproject 1b (SP1b), was one of 11 demonstration systems installed and monitored within the EUPolySMART project [5]. The system employed a TDC driven by district heat from a network supplied by a centralised combined heat and power fired with municipal waste. The system consists of a ClimateWell (4th generation) TDC that pre-cools chilled water for the head office of Borlänge municipality. The conventional chiller has a much greater capacity that the TDC installed in the project, and thus only precooling of the chilled water is possible. The chilled water is used for cooling ventilation air. Fig. 2 shows a schematic of the system including the sensors in the monitoring system. The demonstration system was owned by the utility Borlänge Energi and has been designed by the consulting company Ångpannaföreing (ÅF) subcontracted by ClimateWell AB. The site was identified at the end of 2006 and installation commenced during the late spring of 2007, with hydraulics being completed in July 2007, when the initial testing was started. The system was tested during the rest of the cooling season in 2007 and 2008. As the TDC has been employed to just pre-cool the chilled water return, the
2
3
3314
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Fig. 2. Schematic of the demonstration system, including monitoring sensors in three different circuits.
expressed simply through standard equation. Please note that the values mentioned below are based on energies as charge/discharge cycles are independent and instantaneous values are not relevant. COPth,system = COPth,TDC =
Q
Cdn
Q
TdcCc
COPel,sys = (1)
QCHP
(2)
QTdcDc
QCdn is cold produced by the system (measured), QCHP is the driving heat from district heating (system), QTdcCC is cold produced by the TDC system (measured) and QTdc,Dc is the heat supplied to the TDC system. A calculation of the electrical COP of the TDC circuit takes into account the pressure drop in the heat exchangers. These include the electrical consumption in TDC alone plus electric consumption to overcome the internal pressure drop of the machine in the internal hydraulic circuits. COPel,TDC =
QTdcCc ETdc +
3
(3)
E i=1 dpi
ETdc is the electricity consumption of TDC alone (measured), Edpi is estimated electricity consumption due to the internal pressure drop of the TDC in the three hydraulic circuits (estimated) and Edpi is estimated according to; Edpi =
˙ i · pi m · ti
on the application and is not an inherent component of the cold production system. However, a general formula is given below:
(4)
˙ i is the flow rate of circuit i (i = driving circuit, heat rejection cirm cuit, cooling circuit (measured)) [m3 /s], pi is the pressure drops in circuit i at each flow rate (from manufacturers’ data sheet) [Pa], is the electric efficiency of standard pump. 0.3 was used in this study [Why /Wel ]. The ti is the observation/evaluation time when the circuit i is active. This time was derived from the operation time of the respective pump (measured). The electrical COP of the system (COPel,sys ) includes the electric requirements for the TDC (ETdc ), the three hydraulic circuits (Ecircuit ) and the heat rejection unit (EHr ). It is based on the cold provided to the distribution system (QCDN ) and thus includes possible cold losses. But it excludes the distribution system as this depends
ETdc +
n
QCdn
E i=1 circuit,i
+ EHr
(5)
3. Monitoring results and discussion The whole system was set into operation and various operational tests were carried out during August 2007. A number of faults and problems i.e. pumps, logger, control systems, etc. were identified from the testing period. The suggested changes to the system were implemented during the spring of 2008 and tested during the early summer. This resulted in improved performance of the TDC as well as the system as a whole. The system was finally determined as commissioned on 25th June 2008. The system was run with only minor changes for the rest of the cooling season, which finished on 5th September. After this date, the average outside ambient temperature during the 06:00–17:00 period set for cooling was not above 12.7 ◦ C. Measured data were obtained via a data logger with sensors for temperature, flow and electrical power. Sensors are shown in Fig. 2. Heat powers were calculated from the temperature and flow measurements. Electrical meters were connected to measure the instantaneous power and energies are derived by time integration. Measurements were made every 10 s, but average values for each minute were stored. 3.1. Example day Fig. 3 shows data for a good example of the overall operation of the system on a hot day. At the start of the day there is no activity as the TDC can only deliver cooling when the main compressor chiller is in operation (06:00–17:00). The chilled water starts circulating at 06:00 and the main chiller is operating intermittently at lower power during the early morning before operating continuously until the afternoon when it then operates intermittently at higher power when the outside ambient exceeds 25 ◦ C. The TDC attempts to start providing cooling at the start of this period (green spikes during the morning in the lower diagram), but the
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
3315
Fig. 3. Plot of temperatures (upper diagram) and heat transfer rates (lower diagram) for the TDC for 3rd July 2008. Red for driving circuit (Dc), green for recooler (Rc) and blue for chilled water (Cc), grey for outside ambient temperature (OA). For the compressor chiller the electric power is given (grey). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
temperature of the reactor is still too low for operation of the TDC and it shuts down after a short time. TDC operation starts at around 10:40 with both charging and discharging (cooling). At ∼13:50 charging stops when that barrel becomes fully charged. At ∼14:30 the barrel providing cooling is fully discharged and the barrels are “swapped”, so that the barrel that was providing cooling is then charged and vice versa. During this period cooling as with the cooling capacity, the charging power reduces during the charging cycle (swap) (about 15 min). The return temperature to the TDC from the chilled water network, after the heat exchanger, is 13–17 ◦ C depending on the operation of the main chiller that only has two levels of cooling power. The TDC provides cooling at full capacity, but with these operating conditions it can only provide at best 7 kW of cooling power. The cooling capacity reduces steadily during the discharge process and the TDC can only provide 2–3 kW at the end of this period. For driving circuit, the driving temperature from the district heat (after the heat exchanger) is normally around 75 ◦ C, but occasionally goes
up above 80 ◦ C when hotter “plugs” come from one of the heat sources in the network. At the start of the charging process (around 14:30–15:00) the return temperature is relatively low, but for the rest of the charging process the temperature drop across the TDC is 5–10 ◦ C. In recooling circuit, the recooler fan is operated with an on/off controller at set temperatures of 26/28 ◦ C but the return temperature from this unit (green) oscillates between 26 and 30 ◦ C, due to different placement of controller and monitoring sensors. After ∼12:00 it is on continuously and at 17 kW recooling power has a return temperature ∼6 ◦ C above outside ambient. 3.2. Cooling season 2008 Within cooling season, the period 21st July to 1st August is the period when the system was working properly and when there was hot weather. Data for this period, together with the whole operation period after commissioning are presented in Tables 3 and 4 and Fig. 4. These show that the thermal COP for the system is less
Fig. 4. Breakdown of electrical use for the system during whole operation period (left) and during hottest period (right).
3316
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Table 2 General boundary conditions of the demonstration plant and components.
Table 4 Running times and number of starts for the TDC and for the main compression chiller.
System
Description
Period
CHP
Municipal waste fired CHP plant produces electricity and heat that is distributed via district heating system. The TDC system is connected to district heating via a heat exchanger. Flow from the district heat network is turned on via a solenoid valve when the TDC is being charged. There is no other flow control. The average supply temperature during the monitoring period was 77.7 ◦ C. The TDC is a 4th generation ClimateWell from 2007. The nominal cooling capacity of the TDC is 10 kW, but for the given boundary conditions of the system it is only 7 kW. The heat rejection for the system is with a dry cooler (Flexcoil VTHD) with design capacity of 25 kW at 27 ◦ C ambient temperature with 35/30 ◦ C fluid temperatures at a flow rate of 0.5 kg/s. The dry cooler fan (560 W) is controlled on/off to maintain a return temperature of 26 ◦ C (hysteresis 2 ◦ C). The TDC has integral storage of 25 kWh cool per barrel and so there is no additional storage in the system. The building served by this system is the city hall for the municipality of Borlänge, Sweden. The building, built in 1976 to the then current building regulations, comprises six wings with a total floor area of 15,600 m2 . Approximately 330 people work there in an office environment. The total annual heat load is approximately 1600 MWh and electricity demand 1700 MWh. No statistics are available for the cooling load. Active cooling is only required when the ambient temperature is above ∼13 ◦ C. Cooling is supplied via air handling units, one for each of the six wings, using chilled water supplied from the central chiller system, located in the same technical room as the district heat substation. The cooling distribution system is designed for 12 ◦ C supply temperature and is in operation between 06:00 and 17:00 weekdays. The studied system is an addition to this existing system and is connected to the cooling supply via a heat exchanger, the heating system being unaffected. The TDC is only used for cooling. The return temperature from the cold distribution system varies dependent on the load (outdoor ambient temperature) and operation of the main compression chiller. Cooling season is generally from mid-May to mid-September.
Normal operation (25th June – 5th September) – 321 Main chiller TDC 69 221 Hottest period (21st July – 1st August) – 99 Main chiller 26 82 TDC
TDC
Heat rejection
Storage
Load
Cold distribution system
than 0.30 for the whole operational period and 0.38 for the hot period. The corresponding figures for the TDC only are 0.41 and 0.50, which is roughly 12 percentage points higher. This is due to the fact that there is a significant heat loss within the system for the heat exchangers and pipes between the district heat supply and the TDC and between the TDC and the chilled water delivery circuit. In the case of the chilled water there are also two pumps which add thermal energy to the chilled water thereby reducing the cooling power delivered. For both these circuits roughly 14–17% of the
Num. starts [–]
Delivery [h]
Recooler fan [h] 321 543 99 267
energy is “lost” between the supply and the TDC, resulting in significantly worse COP values for the system compared to only the TDC. For the electrical COP, the figures are really low, being only 1.0 and 1.46 for the system COP over the whole period and hot period respectively. For the TDC only, the values are 2.10 and 4.57. Fig. 4 shows the breakdown of the electrical use for these two periods. For the whole operational period nearly half the electricity is used to run the TDC, while it is only a quarter during the hot period, the recooler fan becoming significantly more important. This highlights three important features of electrical use in the system: (1) The TDC has two barrels, and in each of these barrels there are two internal pumps, one each for the water and solution circulation. The ones for the solution circulation are on continually, whether the TDC is in operation or not. Thus for periods with relatively continuous operation, the electrical use of the TDC is relatively low compared to the thermal energies. Conversely for periods with little operation, the electrical use of the TDC is relatively large. (2) The pumps in the external circuits require a significant amount of energy. There are four pumps, with a total power of ∼500 W. As the delivered cooling power from the TDC is only 2–7 kW, this pump energy is very significant. (3) The recooler fan is controlled on/off with a 2 ◦ C set hysteresis. For hot outdoor ambient temperatures, this fan is on continuously at ∼600 W. For other periods, when ambient temperature is roughly below 22 ◦ C, the fan is in on/off operation and the average electrical use is significantly lower. The running times of the main compression chiller and the TDC show significant differences. For the hot period the availability of the TDC (for cooling) is only 82% and is even less for the whole period (see Table 3), which can be partially explained by the two week period when the TDC was not operating due to a burnt out wire. There are two other causes of the reduced availability of the TDC: (1) The TDC needs to swap between barrels at the end of each charge/discharge cycle, and for ∼15 min cannot deliver cooling. (2) The TDC often starts relatively late during the day (see Fig. 3). This was due to the TDC controller algorithm to avoid unwanted crystallization. An improved algorithm was developed, allowing an earlier state, but not tested during this monitoring period.
Table 3 Energy and COP key figures for the demonstration system and for the TDC itself. Period
COPth [–]
COPel [–]
QCHP [kWh]
Normal operation (25th June–5th September) 0.29 1.00 2034 System 0.41 2.10 1696 TDC Hottest period (21st July–1st August) 1.46 769 System 0.38 4.57 699 TDC 0.50 a b
QCdn cooling [kWh]
Qrecool [kWh]a
596 691
2424 2339
597 280
13%
17%
292 337
989 1008
200 52
12%
14%
The recooling energy was measured only at the outlet of the TDC and not at the dry cooler. The electrical consumption for the TDC is for only the TDC and not external circuits or recooler.
EBOP+Tdc [kWh]b
Loss-DH [%]
Loss-CDN [%]
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
The operation during the cooling season of 2008 has revealed a number of weaknesses in the system and several of them can be improved. The main problem to be addressed is the high use of electricity to run the system. Running time for cool delivery is less than half of that of the recooler fan. The recooler circuit is in operation for longer than the cool delivery because it is required also during charging, which is asynchronous with cool delivery. The charge times are long due to lower power towards end of cycle. As it is now, it makes no sense to have the system as the electrical COP is lower than that of the main compression chiller. A significant reduction of the electrical use could be achieved by replacing the current TDC with the newer version that has no internal pumps as well as reduced pressure drops in the heat exchangers. This and other improvements aimed to increase electrical COP were studied using simulations, based on the model described later in this article. These simulations are reported in system simulation and parametric study [13].
Table 5 Details of the system component. Detailed model description can be found in [14], unless otherwise stated. System component
Description
Centralised CHP model
Simplified model: Type 5 with constant inlet temperatures and flow on hot (district heat) side. The district heat supply is not modelled explicitly, rather a constant temperature of 77.7 ◦ C, the measured average for the monitoring period, is supplied to the heat exchanger in driving circuit. The flow rate on the district heat side is also constant with the measured flow of 941 kg/h during charging of the TDC. At other times it is zero. Detailed model: Types 215 and 216 for 4th generation TDC [15] The TDC is modelled as a grey box model using two different TRNSYS components: Type 215 for the so called barrel containing reactor, evaporator/condenser, and storages for water and LiCl solution; and Type 216 for the switching unit/controller. In a complete TDC there are two barrels and a switching unit, and this is reflected in the TDC model. The models are described in more detail by Bales and Ayadi [15], who also show parameter values and the validation results based on lab measurements. All parameter values used in this study are the same apart from the UA-values for heat transfer for charge/discharge and for losses to ambient. These were adjusted in order to give good agreement with the monitoring data from SP1b. The TDC has cold storage as an integral part of each barrel, and no other storage is included in the system model. Simplified model: Two pipes (Type 31) in each circuit (flow and return) together with a counter flow heat exchanger (Type 5) with constant UA-value. Two pipes are included in each circuit. The size and approximate UA-value for losses were estimated based on the data for the SP1b system, but UA-values were adjusted in order to give good agreement with the monitored data. This was done for each circuit independently using data from quasi steady state operation. The size (thus dynamics) was not adjusted based on monitored data. Cold distribution was not modelled explicitly, rather the flow (FCdn ) was a constant 1116 kg/h during operation (when TOA was above 13 ◦ C during 06:00–17:00 on office days) and the return temperature from the cold distribution (TCdnFl ) used the following correlation derived from monitored data (see Section 4.2). TCdnFl = (13.0 − 0.5 × LT(TOA , 20.0) × (20 − TOA )/7 + GE(TOA , 20.0) × (TOA − 20)) Detailed model: Type 52 for dry cooler. Type 52 was used for modelling the dry cooler. Parameters were derived based on the manufacturer’s data and then adjusted to fit with the monitored data together with the heat loss coefficient of the connecting pipes. Detailed model: Type 23 A PID regulator of Type 23 was used to maintain a constant return water temperature of 27 ◦ C from the dry cooler by varying the air mass flow. This approximated the action of the real on/off control at 26/28 ◦ C. Forcing functions: Detailed model: Type 14
TDC and cold storage
4. Calibration of system model Calibration of system model has been made by comparing simulated results with measured data obtained during cooling season of 2008 (see Table 1). 4.1. Description and modelling of system components The complete TRNSYS system model is shown in Fig. 5, not including the output components. Details description of the system components are given in Table 5.
Pipes and cold distribution
4.2. Calibration of subsystem models Calibration of the subsystem models was performed for driving circuit (Dc), cooling circuit (Cc) and recooling circuit (Rc). The subsystem is calibrated first to define input parameters, heat losses in the pipes and overall heat transfer coefficient (UA-values) of heat exchangers as well as air mass flow in the dry cooler. These subsystem components were simulated and calibrated until the simulation results i.e. temperatures, power and heat transfer rate are similar to measured data (within 4% for power quantities). The measured data (steady state points of operation) of each subsystem were taken randomly; but to give a good spread of data points. Once calibration of subsystems was completed and all parameters were defined, the calibrated parameters and subsystems were transferred to a system model (see Fig. 5). The interactive process and general concept of subsystem calibration is shown in Fig. 6. For the dry cooler, the monitored electrical use as well as inlet and outlet conditions for the water loop and the inlet conditions for the air were used in a model of the recooling circuit. The PID controller was used to control the air mass flow to give the monitored return temperature. The simulated air mass flow was then correlated with measured average fan power (see Fig. 7). The correlation shows that there is a small air flow even with no fan power, which represents the natural convection that can occur in such circumstances. The cold distribution was not modelled explicitly. Instead a function describing the return temperature from the distribution circuit was derived based on the measured data for 5 days of operation. The return temperature (TCdnFl ) is plotted against the outside ambient temperature in Fig. 8. The different colours and symbols are for different days of operation. It can be seen that there is a large scatter in the data. This is mainly due to the operation of the large compression chiller that has three stages of modulation. TCdnFl decreases in a step change when the chiller increases a modulation stage. The black line shows the correlation that was derived.
3317
Heat rejection
PID controller
Components for system control
Electrical model: TDC, pumps and fans
Two forcing functions together with an equation were used to determine when cooling was possible. These limited cooling to be able to occur during the period 06:00–17:00 on office days (no holidays were implemented as the cooling system was on during the whole summer). Additionally cooling is only turned on when the outside ambient temperature is above 13 ◦ C. Detailed model: No specific types. Implemented in equations and as parameters in existing models. The power of the pumps was estimated based on the nominal powers, the measured flows, the pump curves and the measured electrical energy use, which was for all pumps together and not individually. These values were then applied as the pump power in each of the pumps of Type 3 in the system model.
3318
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Fig. 5. TRNSYS studio representation of base case system model with main subsystems marked.
It is essentially in the middle of the scatter for ambient temperatures above 20 ◦ C. TDC was not calibrated as a subsystem, rather the model was derived from lab measurements [15] and implemented directly into the system model. 4.3. Calibration of complete system model The complete system model was calibrated using a five days monitoring period with relatively high cooling loads. The charge/discharge cycles were controlled by the TDC controller
Fig. 6. Basic methodology for calibration of subsystem models.
model and the start conditions (state of charge and temperatures) of the TDC were adjusted to be as close to those in reality as possible by estimating the water content in the internal TDC water store from the monitored data from the TDC controller. This means that the TDC in the model is not charging/discharging in phase with the monitored data and that the time points for the swaps between charge and discharge are not at the same time.
Fig. 7. Correlation of fan power with air mass flow in dry cooler based on simulation of PID controller controlling water return temperature to 27 ◦ C.
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
3319
Table 6 Summary of energy performance figures for the calibration period together with relative differences.
Meas Sim % Diff
QDh [kWh]
QCdn [kWh]
QTdcDc [kWh]
QTdcCc [kWh]
QTdcRc [kWh]
Epump [kWh]
ETdc [kWh]
EHr [kWh]
EBOP+Tdc [kWh]
395.5 397.5 0.5%
162.8 163.2 0.2%
352.0 356.2 1.2%
184.4 185.3 0.4%
517.8 519.6 0.3%
41.1 40.9 −0.5%
19.9 20.0 0.4%
27.9 27.4 −1.9%
89.0 88.3 −0.7%
Inputs (from monitoring data on a 1 min time scale): • Outside ambient conditions (TOA and RHOA ). • Flow temperature from the district heating network (TDhFl ). The . flow rate (mDh ) was assumed constant at 941 kg/h. • Return temperature from cold distribution loop (TCdnFl ). The flow . rate (mCdn ) was not measured; instead it was calculated from the energy balance of the Cc heat exchanger using the monitored temperatures and flows and assuming zero heat losses/gains. • Constant indoor ambient temperature of 24.5 ◦ C (average for the five day period). The system was then simulated and the following parameters were adjusted in order to get good agreement between the summed thermal and electrical energy quantities at system and TDC level: • Changes in the ClimateWell Chiller identified parameters: ◦ Heat loss coefficients for the reactor, condenser/evaporator (including switching unit) as well as solution and water stores was reduced to 2.2 W/K from 8.6 W/K and those for the solution and water stores from 9.2 to 2.2 W/K. ◦ Heat transfer coefficient during charge was reduced to 3330 from 10,000 W/K. ◦ Heat transfer coefficient during discharge was increased from 4230 to 5560 W/K. • The heat loss coefficient for the pipes in the Dc was increased from 2.5 to 8.3 W/m2 K. • The TDC internal power was increased during standby as the real power is not constant. One of the internal pumps is turned on and off occasionally. • Pump power was reduced by 50 W when the TDC was discharging but not charging. This is in accordance with the monitored data, but as only the total pump power was monitored it could not be fully explained. It is likely to be due to the fact that Rc loop had lower flow rate as it flows only through the discharging reactor heat exchanger and not through the charging condenser’s heat exchanger as well.
Fig. 8. Return temperature of the cold delivery loop (TCdnFl ) plotted against the outside ambient temperature (TOA ) for five operation days. The different colours of markers are for different days of operation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
Table 7 Summary of results for thermal and electrical COP at both system and TDC level for the calibration period together with relative differences. Also shown are the running times for the pumps. System
Meas Sim %Diff
TDC
Simulated
COPth
COPel
COPth
COPel
DcPump
CcPump
RcPump
0.41 0.41 −0.3%
1.83 1.85 1.0%
0.52 0.52 −0.7%
2.07 2.10 1.2%
93.9 90.4 −3.7%
39.0 39.7 1.8%
111.7 109.5 −2.0%
Figs. 9 and 10 show the time plots for thermal power (QDh and QCdn ) and electrical power respectively. The results show that there is good agreement in terms of levels, but that the time point for swaps is different. The latter indicates that the controller model is not completely accurate, but the focus of the calibration was on agreement of energy quantities over the whole calibration period of five days, rather than detailed modelling of dynamics.
Fig. 9. Time plot of QCdn and QDh for both monitored data (thin line, red and pink respectively) and simulated values (thick line, blue and green respectively) for one day. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
Fig. 10. Time plot of sum of pump power (thin line: red measured, thick line: blue simulated) as well as TDC power (pink measured – thin line and green simulated – thick line) for one day. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
3320
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Table 8 Summary of energy performance figures for the calibration period together with relative differences. The simulation uses TRNSYS weather data and derived correlation for TCdnFl .
Meas Sim % Diff
QDh [kWh]
QCdn [kWh]
QTdcDc [kWh]
QTdcCc [kWh]
QTdcRc [kWh]
EPump [kWh]
ETdc [kWh]
EHr [kWh]
EBOP+Tdc [kWh]
395.5 389.8 −1.4%
162.8 167.4 2.8%
352.0 347.9 −1.2%
184.4 190.3 3.2%
517.8 515.1 −0.5%
41.1 40.7 −0.9%
19.9 19.6 −1.5%
27.9 24.9 −10.9%
89.0 85.2 −4.2%
The power matching is quite good, apart from for the pump power when the cooling is off in the early afternoon. The data indicates that the Cc pump is still on but that the flow is blocked by an internal valve in the TDC. This is a non-optimised operation of the real controller, and was thus not included in the simulation model. Additionally the TDC has the second water pump running at times when it is not modelled. However, these result in very small electrical use. The identification/calibration results are detailed in Tables 6 and 7. All simulated quantities are within 3% of the measured values apart from the operating time of the Dc pump, which is 4% longer on in reality. A final check of the system model was made by using the TRNSYS weather data for Borlänge [14] and picking out a similar period of weather data as for the calibration period. No inputs from the monitored data were used in the model. The performance figures for this simulation are compared with those from the monitored data in Table 8. This shows that there is a good agreement between the values, although the real boundary conditions were not exactly the same. The average ambient temperature for the simulation was 19.7 ◦ C compared to the monitored value of 20.9 ◦ C. This is the presumed reason for the fan power being significantly lower in the simulation compared to the monitored data.
5. Conclusions The operation during the cooling season of 2008 has revealed a number of weaknesses in the system design and operation. Although a number of weaknesses have been found, it should be stated that the TDC has worked reliably during the whole season. The main problem to be addressed is the high use of electricity to run the system. As it is now, the system provides a relatively low electrical COP and it is lower than that of the main compression chiller. The TDC system is capable of providing maximum thermal and electrical COP’s during the hottest period of around 0.50 and 4.6 respectively. However, the figures were only 0.41 and 2.1 for the complete monitoring period during the summer of 2008 respectively. The lower figures were due to continuous pump operation inside the TDC even during periods of no cold production as well as a period when no cold was produced due to an electrical fault in the overall controller. The figures for the complete system were 0.38 and 1.46 for the thermal and electrical COP respectively for the hot period and 0.29 and 1.0 for the whole summer. This shows significant heat losses between TDC and supply as well as high electricity use compared to cold production. The internal pumps in the TDC and high pump power in the connected circuits accounted for the majority of this energy. System simulation and parametric study [13] of this study determines how this can be reduced. There are a number of other causes of the relatively low thermal and electrical COP values of the entire system such as: • Heat exchanger in the driving circuit causes extra heat losses in the driving circuit of the TDC that significantly reduce the thermal COP of the system compared to that of the TDC. The heat exchanger results in lower operating temperatures as well. The ideal would be to have no extra heat exchanger.
• Heat exchanger in the chilled water delivery circuit causes heat gains in the delivery circuit both through normal gains through the insulation and components but also due to thermal energy from the two pumps used. In addition it forces the TDC to work at lower temperatures, reducing its cooling capacity. The ideal would be to have no extra heat exchanger. • The driving temperature available from the district heating network is lower than ideal for the TDC. It is on average 75–80 ◦ C, whereas 80–90 ◦ C would be more ideal. This affects the charging time of the TDC and thus also parasitic electrical use. • The operating times of the whole system are relatively short. The main compressor chiller is only operating on average 12–27% of the time for the months of the cooling season. This affects the economy as well as the electrical use of the TDC. • The TDC cannot deliver at full power with the normal operating conditions of the system. The system simulation model was calibrated in a three stage process against the monitored data. The final calibration stage showed very good agreement with the monitored data for a five day period with higher cooling loads in terms of both thermal and electrical energy quantities as well as running times of the pumps. The calibrated model is further used for parametric studies in order to find improved system design and control. The simulation study and parametric study of decentralised cooling in district heating network is described in [13]. Acknowledgments The authors wish to convey their sincere appreciation to European Commission (EC) for financial support of this project through FP6-2004-TREN-3 (Contract No. 019988), and similarly to the Swedish Energy Agency for funding via project: P22374-1. References [1] S. Lindmark, The role of absorption cooling for reaching sustainable energy systems, Licentiate Thesis, Division of Energy Processes, Royal Institute of Technology (KTH), Sweden, 2005. [2] M. Rydstrand, Heat-driven cooling in district energy systems, Licentiate Thesis, Division of Energy Processes, Royal Institute of Technology (KTH), Sweden, 2004. [3] L. Trygg, S. Amiri, European perspective on absorption cooling in a combined heat and power system—a case study of energy utility and industries in Sweden, Applied Energy 84 (2007) 1319–1337. [4] A. Egeskog, J. Hansson, G. Berndes, S. Werner, Co-generation of biofuels for transportation and heat for district heating systems-an assessment of the national possibilities in the EU, Energy Policy 37 (2009) 5260–5272. [5] PolySMART, EU-Integrated Project, Project No. 019988, FP6-2004-TREN-3. POLYgeneration with advanced Small and Medium scale thermally driven Airconditioning and Refrigeration Technology (PolySMART). www.polysmart.org. [6] ClimateWell AB, Product description – ClimateWellTM 10 (4th generation), Available at http://www.thermostat.com.tw/pdf/cognition 10.pdf (accessed 6.05.10). [7] M.R. Conde, Properties of aqueous solution of lithium and calcium chlorides: formulations for use in air conditioning equipment design, International Journal of Thermal Sciences 43 (2004) 367–382. [8] R. Olsson, M. Kaarebring-Olsson, S. Jonsson, A Chemical Heat Pump, Patent No: WO0037864, Sweden, 2000. [9] R. Boer, W.G. Haije, J.B.J. Veldhuis, Determination of structural, thermodynamic and phase properties in the Na2 S–H2 O system for application in a chemical heat pump, Journal of Thermochimica Acta 395 (2003) 3–19.
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321 [10] H. Ogura, T. Yamamoto, H. Kage, Efficiencies of CaO/H2 O/Ca(OH)2 chemical heat pump for heat storing and heating/cooling, Energy 28 (2003) 1479– 1493. [11] C. Bales, S. Nordlander, TCA Evaluation, Lab Measurements, Modelling and System Simulations. No. 3-9809656-4-3, Högskolan Dalarna, Borlänge, Sweden, 2005, ISSN 1401-7555. [12] F. Holmberg, Alternative Refrigerants for ClimateWell’s Sorption Cooling Technology, Master Thesis, Division of Energy Processes, Department of Chemical Technology, Royal Institute of Technology (KTH), Sweden, 2008.
3321
[13] S. Udomsri, C. Bales, A.R. Martin, V. Martin, Decentralised cooling in district heating network: system simulation and parametric study, Applied Energy, submitted for publication, Manuscript Number: APEN-D-10-01610. [14] TRNSYS, TRNSYS Coordinator, Solar Energy Laboratory, University of Wisconsin-Madison, 1500 Engineering Drive, 1303 Engineering Research Building Madison, WI 53706 – U.S.A. [15] C. Bales, O. Ayadi, Modelling of commercial absorption heat pump with integral storage, in: Proceedings Effstock 2009, CD-rom, Swedvac, Stockholm, Sweden, 2009.
Contents lists available at ScienceDirect
Energy and Buildings journal homepage: www.elsevier.com/locate/enbuild
Decentralised cooling in district heating network: Monitoring results and calibration of simulation model Seksan Udomsri a,∗ , Chris Bales b , Andrew R. Martin a , Viktoria Martin a a b
Division of Heat and Power Technology, Department of Energy Technology, Royal Institute of Technology (KTH), 100 44 Stockholm, Sweden Solar Energy Research Center (SERC), Högskolan Dalarna, 781 88 Borlänge, Sweden
a r t i c l e
i n f o
Article history: Received 17 December 2010 Accepted 1 August 2011 Keywords: Thermally driven chillers Decentralised cooling District heating network Monitoring results Calibration of simulation model TRNSYS
a b s t r a c t This article presents the monitoring results of a thermally driven chiller (TDC) driven by district heat from a network supplied by a centralised combined heat and power (CHP) fired with municipal waste. The main objective of this article is to analyse the monitoring results obtained from the demonstration and calibrate a system model that is later used for parametric studies in order to find improved system design and control. The calibration of the system model was made in three stages and all the energy performance figures were within 4% of the measured values. Results show that the TDC system is capable of providing maximum thermal and electrical COP’s of 0.50 and 4.6 respectively during the hottest period. For the complete monitoring period during the summer of 2008, the figures were 0.41 and 2.1. The lower figures were due to continuous pump operation inside the TDC even during periods of no cold production and a period when no cold was produced. However the internal pumps inside the TDC have been removed in the new version TDC to increase the electrical COP. System simulation and parametric studies will be employed to further determine how the electrical COP can be improved. © 2011 Elsevier B.V. All rights reserved.
1. Introduction 1.1. Background and objectives The demand of comfort cooling has dramatically increased in the EU over the past few years, even in cold climate like Sweden, and district cooling market is expected to expand tremendously in the next decade and beyond. In Sweden, the demand for cooling has increased over the past few years despite its upper northern latitude [1,2]. However, the largest cooling demand has always been found in summer period where low or no heating load exists. Efforts are being made to enhance thermal power plants via absorption technologies in order to increase the exploitation of heat and utilize low-grade heat for cooling production. Various alternative sources of cooling production have been explored intensively to replace mechanical chillers towards reducing electrical energy consumption and CO2 emissions [1–4]. One concrete example of particular significance is decentralised cooling in district heating networks. It is an attractive approach where the cooling is provided on the demand side and consumers can deploy the system themselves to produce cooling by using heat from district heating network.
∗ Corresponding author. Tel.: +46 8 790 6140; fax: +46 8 20 41 61. E-mail addresses: [email protected], [email protected] (S. Udomsri). 0378-7788/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.enbuild.2011.08.001
This article presents the monitoring results and calibration simulation of the demonstration system of decentralised cooling in district heating network, which is one of the 11 demonstration systems installed and monitored within the EU-PolySMART project (www.polysmart.org) [5]. The system employs a ClimateWell [6] thermally driven chiller (TDC) supplied by heat from the district heating system. The PolySMART project aims were to develop a set of technical solutions for a new market segment of polygeneration, in particular the market for small tri-generation systems (combined production of electric power, heat and cooling). The key components of these systems always include the combined heat and power (CHP) plant along with the TDC. There are different approaches of composing the CHP with TDC [5]: (i) centralised CHP and centralised TDC in combination with a heating and cooling network, (ii) centralised CHP and decentralised TDC in combination with a heating network, and (iii) decentralised CHP and decentralised TDC both on the demand side. Each approach can be applied under different conditions depending upon the source of generation, consumption, and applications. However, this article reports on a combination of centralised CHP and decentralised TDC in district heating network. District heating can be described as rational and environmentally friendly method to heat residential and commercial buildings, etc. District heating is very common heating method and available throughout Sweden. The CHP unit generates electricity and heat where the heat can be distributed in term of steam and district heating in the district heating network.
3312
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Nomenclature CHP COP Cp DH E Eff HEP LHV ˙ m P PE Q RH SP T TCA TDC Tim V Dpi
combined heating and power coefficient of performance specific heat capacity district heating electricity consumption efficiency high efficiency pump lower heating value of a fuel mass flow power primary energy thermal energy flow relative humidity subproject temperature Thermo-Chemical Accumulator thermally driven chiller time volume flow pressure drop in each circuit
Subscript Air air BOP balance of plant chilling circuit of TDC Cc Cdn cold distribution system for end-use N Dc driving circuit of TDC el electricity feed line (considering as the one leaving the heat Fl source, and thus the hottest line) high efficiency pump HEP Hr heat rejection system Meas measurement OA outside ambient Pump pump recooling circuit of TDC Rc RH relative humidity Rl return line (considering as the one returning to the heat source, and thus, the coldest line) Sim simulation Sys system Tdc thermal driven cooling thermal energy Th ti time
As often found in Europe, especially Sweden, cogeneration facilities are an important contributor to the overall energy mix, in particular for heat delivery in wintertime. One important feedstock for the CHP unit is municipal solid waste (MSW). However the capacity of district heating is not fully utilized during non-heating period or the summer time. This means the CHP plant operates relatively low capacity and low efficiency at consequently. Thus it is important to find new applications for the heat produced during this period e.g. using heat as energy carrier for distributed smallscale heat driven machines. The commercial TDC usually employs heat from steam, hot water or from a solar thermal system or even directly from combustion gases as a heat source. The main application is comfort cooling or for industrial processes. There are several technologies of absorption chillers commercially available today, e.g. a standard absorption chiller using LiBr/water or NH3 /water and salt-water absorption chiller and/or chemical heat pump. Absorption chillers are more common at medium or larger
scale, while small scale TDC are in process of becoming commercial. Chemical heat pump is new and promising technology in which it is capable of powering its system with low temperature heat sources. Salt-water solutions such as lithium chloride (LiCl)/water, sodium sulphite (Na2 S)/water, and calcium chlorides (CaCl2 )/water, etc. have been used [7–10]. The TDC in this study is a recently commercialised chemical heat pump using LiCl/water as a working fluid pair [8,11]. The objective of the demonstration was to demonstrate the use of the ClimateWell chiller in distributed cooling with centralised CHP in order to develop the best system configuration for the TDC using a particular form of chemical heat pump, the ThermoChemical Accumulator (TCA). The main objective of this article is to present the results obtained from the demonstration as well as to calibrate a system model that was later used for parametric studies in order to find improved system design and control. Results obtained from monitored data will be presented and verified against simulated results obtained via dynamic modelling with TRNSYS (TRaNsient SYstem Simulation program). 1.2. Thermally driven chiller (TDC) A chemical heat pump or Thermo-Chemical Accumulator (TCA), patented in 2000 [8], has been employed and installed in this project as a TDC unit. It has been developed and is sold by a Swedish company ClimateWell AB. It is a three-phase absorption chillers/heat pump that is capable of storing energy internally with high energy density in the form of crystallized salt (LiCl) with water as refrigerant. In principal, the process operates under vacuum condition like in standard absorption chillers using LiBrwater. The TCA is however significantly different from traditional absorption chillers in that it works in batch mode with relatively long cycle times (>6 h). The common heat sources for this are district heating, waste heat from cogeneration and solar thermal collectors. Fig. 1 shows diagram of the ClimateWell chiller including major components that is referred to as the TDC system in the rest of this article. The triple-state process, so called because it uses solid, solution and vapour at the same time, makes it particularly different from other chemical heat pumps or standard absorption processes (two phase) [11]. It consists of two identical units, so called barrels, that work together to provide quasi-continuous operation (see Fig. 1). Each barrel consists of a reactor and a condenser/evaporator that is connected by the gas pipe or vapour channel. The reactor, called generator/absorber in normal absorption chillers, contains sorbent/hygroscopic salt solution (LiCl), while the condenser/evaporator contains pure water (refrigerant). The system operates intermittently with a charge phase followed by discharge phase. In the charging phase, salt-water solution is heated by a thermal source via a heat exchanger in the reactor and the solution becomes steadily more concentrated. This can be continued until solid is formed. The solid (crystals) are physically restrained from being transported from the reactor and causing clogging. At the same time water is evaporated and steam is released to the condenser/evaporator. In discharge process, a reversed process takes place, with a heat exchanger transferring heat from the building to the water, which evaporates and is transported to the reactor. The water gets absorbed by the concentrated salt, either in solid or solution form depending on the state of charge, and heat is released and transferred to an external circuit via a heat exchanger [6,11,12]. The technology has been developed through five different generations, each with its own particularities that make it significantly different in terms of operation from its predecessors. In this study, two versions of the TDC were analysed: the 4th generation as installed in the demonstration system, subproject 1b (SP1b) in Sweden, and a 5th generation as was installed in the Madrid, Spain
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
3313
Fig. 1. Working principal of the TCA unit with major components included [left] and drawing of ClimateWell solar chiller with two barrels [right] [6].
(SP1a). The 5th generation is the one currently sold (2010) is implemented in system simulations and parametric studies of SP1b that are described in Decentralised cooling in district heating network: system simulation and parametric study [13].
2.2. General boundary conditions for the plant
1.3. Methodology
Table 2 shows general boundary conditions and components of the plant, mostly for CHP, TDC and heat rejection unit.
The demonstration system of decentralised cooling driven by district heating network was installed, monitored and calibrated within the course of the project. The calibration simulation model was made and compared again monitored data. Performance figures for evaluation of the results were defined and criteria was set that the model should predict the use and delivered energy quantities of within 4% of measured values. The following steps of system calibration and simulation is presented in Table 1.
evaluated system is thus only a part of the total cooling system and does not operate during the winter.
2.3. Definitions of COP General formula of thermal and electrical COP’s is defined in this section. The thermal COP for both system and TDC unit can be Table 1 Steps of calibration in the system study and simulation. Step
Calibration of system model against monitored data
1
The calibration of the system model was made in three stages: (i) estimation of parameters based on manufacturer data and dimensions of the system; (ii) calibration of each circuit (pipes and heat exchangers) separately using steady state data points; (iii) and finally calibration of the complete model in terms of thermal and electrical energy as well as running times, for a five day time series of data with 1 min average data values. Subsystem calibration. The three subsystems for the driving, recooling and cooling circuits to the TDC were calibrated against the measured data in terms of energy balance using a range of steady state values from different operating states. The UA-values of the heat exchangers and of the pipes are the parameters that were varied in the calibration process. Determination of the power use of the fan in the dry cooler as a function of the air flow. This was carried out based on the identified dry cooler parameters, measured weather data and inlet/outlet temperatures together with average electrical power for quasi steady state periods. The flow temperature from the cooling distribution loop coming from the air handling units (TCdnFl ) was analysed and a simple correlation between TCdnFl and ambient temperature was derived. The complete system model was then calibrated against a five day dynamic measurement sequence from a hot period. As the TDC has significant internal thermal storage, the starting state of charge had to be determined for each of the two internal storage units (barrels). The main criteria for calibration were the thermal and electrical energies of the whole system. The parameters that were varied in order to gain a good fit were the control parameters for TDC, electrical power of components, UA-values for the TDC heat exchangers and losses from the internal stores.
2. Demonstration plant and performance figures 2.1. Description of the demonstration system The demonstration plant, subproject 1b (SP1b), was one of 11 demonstration systems installed and monitored within the EUPolySMART project [5]. The system employed a TDC driven by district heat from a network supplied by a centralised combined heat and power fired with municipal waste. The system consists of a ClimateWell (4th generation) TDC that pre-cools chilled water for the head office of Borlänge municipality. The conventional chiller has a much greater capacity that the TDC installed in the project, and thus only precooling of the chilled water is possible. The chilled water is used for cooling ventilation air. Fig. 2 shows a schematic of the system including the sensors in the monitoring system. The demonstration system was owned by the utility Borlänge Energi and has been designed by the consulting company Ångpannaföreing (ÅF) subcontracted by ClimateWell AB. The site was identified at the end of 2006 and installation commenced during the late spring of 2007, with hydraulics being completed in July 2007, when the initial testing was started. The system was tested during the rest of the cooling season in 2007 and 2008. As the TDC has been employed to just pre-cool the chilled water return, the
2
3
3314
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Fig. 2. Schematic of the demonstration system, including monitoring sensors in three different circuits.
expressed simply through standard equation. Please note that the values mentioned below are based on energies as charge/discharge cycles are independent and instantaneous values are not relevant. COPth,system = COPth,TDC =
Q
Cdn
Q
TdcCc
COPel,sys = (1)
QCHP
(2)
QTdcDc
QCdn is cold produced by the system (measured), QCHP is the driving heat from district heating (system), QTdcCC is cold produced by the TDC system (measured) and QTdc,Dc is the heat supplied to the TDC system. A calculation of the electrical COP of the TDC circuit takes into account the pressure drop in the heat exchangers. These include the electrical consumption in TDC alone plus electric consumption to overcome the internal pressure drop of the machine in the internal hydraulic circuits. COPel,TDC =
QTdcCc ETdc +
3
(3)
E i=1 dpi
ETdc is the electricity consumption of TDC alone (measured), Edpi is estimated electricity consumption due to the internal pressure drop of the TDC in the three hydraulic circuits (estimated) and Edpi is estimated according to; Edpi =
˙ i · pi m · ti
on the application and is not an inherent component of the cold production system. However, a general formula is given below:
(4)
˙ i is the flow rate of circuit i (i = driving circuit, heat rejection cirm cuit, cooling circuit (measured)) [m3 /s], pi is the pressure drops in circuit i at each flow rate (from manufacturers’ data sheet) [Pa], is the electric efficiency of standard pump. 0.3 was used in this study [Why /Wel ]. The ti is the observation/evaluation time when the circuit i is active. This time was derived from the operation time of the respective pump (measured). The electrical COP of the system (COPel,sys ) includes the electric requirements for the TDC (ETdc ), the three hydraulic circuits (Ecircuit ) and the heat rejection unit (EHr ). It is based on the cold provided to the distribution system (QCDN ) and thus includes possible cold losses. But it excludes the distribution system as this depends
ETdc +
n
QCdn
E i=1 circuit,i
+ EHr
(5)
3. Monitoring results and discussion The whole system was set into operation and various operational tests were carried out during August 2007. A number of faults and problems i.e. pumps, logger, control systems, etc. were identified from the testing period. The suggested changes to the system were implemented during the spring of 2008 and tested during the early summer. This resulted in improved performance of the TDC as well as the system as a whole. The system was finally determined as commissioned on 25th June 2008. The system was run with only minor changes for the rest of the cooling season, which finished on 5th September. After this date, the average outside ambient temperature during the 06:00–17:00 period set for cooling was not above 12.7 ◦ C. Measured data were obtained via a data logger with sensors for temperature, flow and electrical power. Sensors are shown in Fig. 2. Heat powers were calculated from the temperature and flow measurements. Electrical meters were connected to measure the instantaneous power and energies are derived by time integration. Measurements were made every 10 s, but average values for each minute were stored. 3.1. Example day Fig. 3 shows data for a good example of the overall operation of the system on a hot day. At the start of the day there is no activity as the TDC can only deliver cooling when the main compressor chiller is in operation (06:00–17:00). The chilled water starts circulating at 06:00 and the main chiller is operating intermittently at lower power during the early morning before operating continuously until the afternoon when it then operates intermittently at higher power when the outside ambient exceeds 25 ◦ C. The TDC attempts to start providing cooling at the start of this period (green spikes during the morning in the lower diagram), but the
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
3315
Fig. 3. Plot of temperatures (upper diagram) and heat transfer rates (lower diagram) for the TDC for 3rd July 2008. Red for driving circuit (Dc), green for recooler (Rc) and blue for chilled water (Cc), grey for outside ambient temperature (OA). For the compressor chiller the electric power is given (grey). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
temperature of the reactor is still too low for operation of the TDC and it shuts down after a short time. TDC operation starts at around 10:40 with both charging and discharging (cooling). At ∼13:50 charging stops when that barrel becomes fully charged. At ∼14:30 the barrel providing cooling is fully discharged and the barrels are “swapped”, so that the barrel that was providing cooling is then charged and vice versa. During this period cooling as with the cooling capacity, the charging power reduces during the charging cycle (swap) (about 15 min). The return temperature to the TDC from the chilled water network, after the heat exchanger, is 13–17 ◦ C depending on the operation of the main chiller that only has two levels of cooling power. The TDC provides cooling at full capacity, but with these operating conditions it can only provide at best 7 kW of cooling power. The cooling capacity reduces steadily during the discharge process and the TDC can only provide 2–3 kW at the end of this period. For driving circuit, the driving temperature from the district heat (after the heat exchanger) is normally around 75 ◦ C, but occasionally goes
up above 80 ◦ C when hotter “plugs” come from one of the heat sources in the network. At the start of the charging process (around 14:30–15:00) the return temperature is relatively low, but for the rest of the charging process the temperature drop across the TDC is 5–10 ◦ C. In recooling circuit, the recooler fan is operated with an on/off controller at set temperatures of 26/28 ◦ C but the return temperature from this unit (green) oscillates between 26 and 30 ◦ C, due to different placement of controller and monitoring sensors. After ∼12:00 it is on continuously and at 17 kW recooling power has a return temperature ∼6 ◦ C above outside ambient. 3.2. Cooling season 2008 Within cooling season, the period 21st July to 1st August is the period when the system was working properly and when there was hot weather. Data for this period, together with the whole operation period after commissioning are presented in Tables 3 and 4 and Fig. 4. These show that the thermal COP for the system is less
Fig. 4. Breakdown of electrical use for the system during whole operation period (left) and during hottest period (right).
3316
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Table 2 General boundary conditions of the demonstration plant and components.
Table 4 Running times and number of starts for the TDC and for the main compression chiller.
System
Description
Period
CHP
Municipal waste fired CHP plant produces electricity and heat that is distributed via district heating system. The TDC system is connected to district heating via a heat exchanger. Flow from the district heat network is turned on via a solenoid valve when the TDC is being charged. There is no other flow control. The average supply temperature during the monitoring period was 77.7 ◦ C. The TDC is a 4th generation ClimateWell from 2007. The nominal cooling capacity of the TDC is 10 kW, but for the given boundary conditions of the system it is only 7 kW. The heat rejection for the system is with a dry cooler (Flexcoil VTHD) with design capacity of 25 kW at 27 ◦ C ambient temperature with 35/30 ◦ C fluid temperatures at a flow rate of 0.5 kg/s. The dry cooler fan (560 W) is controlled on/off to maintain a return temperature of 26 ◦ C (hysteresis 2 ◦ C). The TDC has integral storage of 25 kWh cool per barrel and so there is no additional storage in the system. The building served by this system is the city hall for the municipality of Borlänge, Sweden. The building, built in 1976 to the then current building regulations, comprises six wings with a total floor area of 15,600 m2 . Approximately 330 people work there in an office environment. The total annual heat load is approximately 1600 MWh and electricity demand 1700 MWh. No statistics are available for the cooling load. Active cooling is only required when the ambient temperature is above ∼13 ◦ C. Cooling is supplied via air handling units, one for each of the six wings, using chilled water supplied from the central chiller system, located in the same technical room as the district heat substation. The cooling distribution system is designed for 12 ◦ C supply temperature and is in operation between 06:00 and 17:00 weekdays. The studied system is an addition to this existing system and is connected to the cooling supply via a heat exchanger, the heating system being unaffected. The TDC is only used for cooling. The return temperature from the cold distribution system varies dependent on the load (outdoor ambient temperature) and operation of the main compression chiller. Cooling season is generally from mid-May to mid-September.
Normal operation (25th June – 5th September) – 321 Main chiller TDC 69 221 Hottest period (21st July – 1st August) – 99 Main chiller 26 82 TDC
TDC
Heat rejection
Storage
Load
Cold distribution system
than 0.30 for the whole operational period and 0.38 for the hot period. The corresponding figures for the TDC only are 0.41 and 0.50, which is roughly 12 percentage points higher. This is due to the fact that there is a significant heat loss within the system for the heat exchangers and pipes between the district heat supply and the TDC and between the TDC and the chilled water delivery circuit. In the case of the chilled water there are also two pumps which add thermal energy to the chilled water thereby reducing the cooling power delivered. For both these circuits roughly 14–17% of the
Num. starts [–]
Delivery [h]
Recooler fan [h] 321 543 99 267
energy is “lost” between the supply and the TDC, resulting in significantly worse COP values for the system compared to only the TDC. For the electrical COP, the figures are really low, being only 1.0 and 1.46 for the system COP over the whole period and hot period respectively. For the TDC only, the values are 2.10 and 4.57. Fig. 4 shows the breakdown of the electrical use for these two periods. For the whole operational period nearly half the electricity is used to run the TDC, while it is only a quarter during the hot period, the recooler fan becoming significantly more important. This highlights three important features of electrical use in the system: (1) The TDC has two barrels, and in each of these barrels there are two internal pumps, one each for the water and solution circulation. The ones for the solution circulation are on continually, whether the TDC is in operation or not. Thus for periods with relatively continuous operation, the electrical use of the TDC is relatively low compared to the thermal energies. Conversely for periods with little operation, the electrical use of the TDC is relatively large. (2) The pumps in the external circuits require a significant amount of energy. There are four pumps, with a total power of ∼500 W. As the delivered cooling power from the TDC is only 2–7 kW, this pump energy is very significant. (3) The recooler fan is controlled on/off with a 2 ◦ C set hysteresis. For hot outdoor ambient temperatures, this fan is on continuously at ∼600 W. For other periods, when ambient temperature is roughly below 22 ◦ C, the fan is in on/off operation and the average electrical use is significantly lower. The running times of the main compression chiller and the TDC show significant differences. For the hot period the availability of the TDC (for cooling) is only 82% and is even less for the whole period (see Table 3), which can be partially explained by the two week period when the TDC was not operating due to a burnt out wire. There are two other causes of the reduced availability of the TDC: (1) The TDC needs to swap between barrels at the end of each charge/discharge cycle, and for ∼15 min cannot deliver cooling. (2) The TDC often starts relatively late during the day (see Fig. 3). This was due to the TDC controller algorithm to avoid unwanted crystallization. An improved algorithm was developed, allowing an earlier state, but not tested during this monitoring period.
Table 3 Energy and COP key figures for the demonstration system and for the TDC itself. Period
COPth [–]
COPel [–]
QCHP [kWh]
Normal operation (25th June–5th September) 0.29 1.00 2034 System 0.41 2.10 1696 TDC Hottest period (21st July–1st August) 1.46 769 System 0.38 4.57 699 TDC 0.50 a b
QCdn cooling [kWh]
Qrecool [kWh]a
596 691
2424 2339
597 280
13%
17%
292 337
989 1008
200 52
12%
14%
The recooling energy was measured only at the outlet of the TDC and not at the dry cooler. The electrical consumption for the TDC is for only the TDC and not external circuits or recooler.
EBOP+Tdc [kWh]b
Loss-DH [%]
Loss-CDN [%]
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
The operation during the cooling season of 2008 has revealed a number of weaknesses in the system and several of them can be improved. The main problem to be addressed is the high use of electricity to run the system. Running time for cool delivery is less than half of that of the recooler fan. The recooler circuit is in operation for longer than the cool delivery because it is required also during charging, which is asynchronous with cool delivery. The charge times are long due to lower power towards end of cycle. As it is now, it makes no sense to have the system as the electrical COP is lower than that of the main compression chiller. A significant reduction of the electrical use could be achieved by replacing the current TDC with the newer version that has no internal pumps as well as reduced pressure drops in the heat exchangers. This and other improvements aimed to increase electrical COP were studied using simulations, based on the model described later in this article. These simulations are reported in system simulation and parametric study [13].
Table 5 Details of the system component. Detailed model description can be found in [14], unless otherwise stated. System component
Description
Centralised CHP model
Simplified model: Type 5 with constant inlet temperatures and flow on hot (district heat) side. The district heat supply is not modelled explicitly, rather a constant temperature of 77.7 ◦ C, the measured average for the monitoring period, is supplied to the heat exchanger in driving circuit. The flow rate on the district heat side is also constant with the measured flow of 941 kg/h during charging of the TDC. At other times it is zero. Detailed model: Types 215 and 216 for 4th generation TDC [15] The TDC is modelled as a grey box model using two different TRNSYS components: Type 215 for the so called barrel containing reactor, evaporator/condenser, and storages for water and LiCl solution; and Type 216 for the switching unit/controller. In a complete TDC there are two barrels and a switching unit, and this is reflected in the TDC model. The models are described in more detail by Bales and Ayadi [15], who also show parameter values and the validation results based on lab measurements. All parameter values used in this study are the same apart from the UA-values for heat transfer for charge/discharge and for losses to ambient. These were adjusted in order to give good agreement with the monitoring data from SP1b. The TDC has cold storage as an integral part of each barrel, and no other storage is included in the system model. Simplified model: Two pipes (Type 31) in each circuit (flow and return) together with a counter flow heat exchanger (Type 5) with constant UA-value. Two pipes are included in each circuit. The size and approximate UA-value for losses were estimated based on the data for the SP1b system, but UA-values were adjusted in order to give good agreement with the monitored data. This was done for each circuit independently using data from quasi steady state operation. The size (thus dynamics) was not adjusted based on monitored data. Cold distribution was not modelled explicitly, rather the flow (FCdn ) was a constant 1116 kg/h during operation (when TOA was above 13 ◦ C during 06:00–17:00 on office days) and the return temperature from the cold distribution (TCdnFl ) used the following correlation derived from monitored data (see Section 4.2). TCdnFl = (13.0 − 0.5 × LT(TOA , 20.0) × (20 − TOA )/7 + GE(TOA , 20.0) × (TOA − 20)) Detailed model: Type 52 for dry cooler. Type 52 was used for modelling the dry cooler. Parameters were derived based on the manufacturer’s data and then adjusted to fit with the monitored data together with the heat loss coefficient of the connecting pipes. Detailed model: Type 23 A PID regulator of Type 23 was used to maintain a constant return water temperature of 27 ◦ C from the dry cooler by varying the air mass flow. This approximated the action of the real on/off control at 26/28 ◦ C. Forcing functions: Detailed model: Type 14
TDC and cold storage
4. Calibration of system model Calibration of system model has been made by comparing simulated results with measured data obtained during cooling season of 2008 (see Table 1). 4.1. Description and modelling of system components The complete TRNSYS system model is shown in Fig. 5, not including the output components. Details description of the system components are given in Table 5.
Pipes and cold distribution
4.2. Calibration of subsystem models Calibration of the subsystem models was performed for driving circuit (Dc), cooling circuit (Cc) and recooling circuit (Rc). The subsystem is calibrated first to define input parameters, heat losses in the pipes and overall heat transfer coefficient (UA-values) of heat exchangers as well as air mass flow in the dry cooler. These subsystem components were simulated and calibrated until the simulation results i.e. temperatures, power and heat transfer rate are similar to measured data (within 4% for power quantities). The measured data (steady state points of operation) of each subsystem were taken randomly; but to give a good spread of data points. Once calibration of subsystems was completed and all parameters were defined, the calibrated parameters and subsystems were transferred to a system model (see Fig. 5). The interactive process and general concept of subsystem calibration is shown in Fig. 6. For the dry cooler, the monitored electrical use as well as inlet and outlet conditions for the water loop and the inlet conditions for the air were used in a model of the recooling circuit. The PID controller was used to control the air mass flow to give the monitored return temperature. The simulated air mass flow was then correlated with measured average fan power (see Fig. 7). The correlation shows that there is a small air flow even with no fan power, which represents the natural convection that can occur in such circumstances. The cold distribution was not modelled explicitly. Instead a function describing the return temperature from the distribution circuit was derived based on the measured data for 5 days of operation. The return temperature (TCdnFl ) is plotted against the outside ambient temperature in Fig. 8. The different colours and symbols are for different days of operation. It can be seen that there is a large scatter in the data. This is mainly due to the operation of the large compression chiller that has three stages of modulation. TCdnFl decreases in a step change when the chiller increases a modulation stage. The black line shows the correlation that was derived.
3317
Heat rejection
PID controller
Components for system control
Electrical model: TDC, pumps and fans
Two forcing functions together with an equation were used to determine when cooling was possible. These limited cooling to be able to occur during the period 06:00–17:00 on office days (no holidays were implemented as the cooling system was on during the whole summer). Additionally cooling is only turned on when the outside ambient temperature is above 13 ◦ C. Detailed model: No specific types. Implemented in equations and as parameters in existing models. The power of the pumps was estimated based on the nominal powers, the measured flows, the pump curves and the measured electrical energy use, which was for all pumps together and not individually. These values were then applied as the pump power in each of the pumps of Type 3 in the system model.
3318
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Fig. 5. TRNSYS studio representation of base case system model with main subsystems marked.
It is essentially in the middle of the scatter for ambient temperatures above 20 ◦ C. TDC was not calibrated as a subsystem, rather the model was derived from lab measurements [15] and implemented directly into the system model. 4.3. Calibration of complete system model The complete system model was calibrated using a five days monitoring period with relatively high cooling loads. The charge/discharge cycles were controlled by the TDC controller
Fig. 6. Basic methodology for calibration of subsystem models.
model and the start conditions (state of charge and temperatures) of the TDC were adjusted to be as close to those in reality as possible by estimating the water content in the internal TDC water store from the monitored data from the TDC controller. This means that the TDC in the model is not charging/discharging in phase with the monitored data and that the time points for the swaps between charge and discharge are not at the same time.
Fig. 7. Correlation of fan power with air mass flow in dry cooler based on simulation of PID controller controlling water return temperature to 27 ◦ C.
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
3319
Table 6 Summary of energy performance figures for the calibration period together with relative differences.
Meas Sim % Diff
QDh [kWh]
QCdn [kWh]
QTdcDc [kWh]
QTdcCc [kWh]
QTdcRc [kWh]
Epump [kWh]
ETdc [kWh]
EHr [kWh]
EBOP+Tdc [kWh]
395.5 397.5 0.5%
162.8 163.2 0.2%
352.0 356.2 1.2%
184.4 185.3 0.4%
517.8 519.6 0.3%
41.1 40.9 −0.5%
19.9 20.0 0.4%
27.9 27.4 −1.9%
89.0 88.3 −0.7%
Inputs (from monitoring data on a 1 min time scale): • Outside ambient conditions (TOA and RHOA ). • Flow temperature from the district heating network (TDhFl ). The . flow rate (mDh ) was assumed constant at 941 kg/h. • Return temperature from cold distribution loop (TCdnFl ). The flow . rate (mCdn ) was not measured; instead it was calculated from the energy balance of the Cc heat exchanger using the monitored temperatures and flows and assuming zero heat losses/gains. • Constant indoor ambient temperature of 24.5 ◦ C (average for the five day period). The system was then simulated and the following parameters were adjusted in order to get good agreement between the summed thermal and electrical energy quantities at system and TDC level: • Changes in the ClimateWell Chiller identified parameters: ◦ Heat loss coefficients for the reactor, condenser/evaporator (including switching unit) as well as solution and water stores was reduced to 2.2 W/K from 8.6 W/K and those for the solution and water stores from 9.2 to 2.2 W/K. ◦ Heat transfer coefficient during charge was reduced to 3330 from 10,000 W/K. ◦ Heat transfer coefficient during discharge was increased from 4230 to 5560 W/K. • The heat loss coefficient for the pipes in the Dc was increased from 2.5 to 8.3 W/m2 K. • The TDC internal power was increased during standby as the real power is not constant. One of the internal pumps is turned on and off occasionally. • Pump power was reduced by 50 W when the TDC was discharging but not charging. This is in accordance with the monitored data, but as only the total pump power was monitored it could not be fully explained. It is likely to be due to the fact that Rc loop had lower flow rate as it flows only through the discharging reactor heat exchanger and not through the charging condenser’s heat exchanger as well.
Fig. 8. Return temperature of the cold delivery loop (TCdnFl ) plotted against the outside ambient temperature (TOA ) for five operation days. The different colours of markers are for different days of operation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
Table 7 Summary of results for thermal and electrical COP at both system and TDC level for the calibration period together with relative differences. Also shown are the running times for the pumps. System
Meas Sim %Diff
TDC
Simulated
COPth
COPel
COPth
COPel
DcPump
CcPump
RcPump
0.41 0.41 −0.3%
1.83 1.85 1.0%
0.52 0.52 −0.7%
2.07 2.10 1.2%
93.9 90.4 −3.7%
39.0 39.7 1.8%
111.7 109.5 −2.0%
Figs. 9 and 10 show the time plots for thermal power (QDh and QCdn ) and electrical power respectively. The results show that there is good agreement in terms of levels, but that the time point for swaps is different. The latter indicates that the controller model is not completely accurate, but the focus of the calibration was on agreement of energy quantities over the whole calibration period of five days, rather than detailed modelling of dynamics.
Fig. 9. Time plot of QCdn and QDh for both monitored data (thin line, red and pink respectively) and simulated values (thick line, blue and green respectively) for one day. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
Fig. 10. Time plot of sum of pump power (thin line: red measured, thick line: blue simulated) as well as TDC power (pink measured – thin line and green simulated – thick line) for one day. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
3320
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321
Table 8 Summary of energy performance figures for the calibration period together with relative differences. The simulation uses TRNSYS weather data and derived correlation for TCdnFl .
Meas Sim % Diff
QDh [kWh]
QCdn [kWh]
QTdcDc [kWh]
QTdcCc [kWh]
QTdcRc [kWh]
EPump [kWh]
ETdc [kWh]
EHr [kWh]
EBOP+Tdc [kWh]
395.5 389.8 −1.4%
162.8 167.4 2.8%
352.0 347.9 −1.2%
184.4 190.3 3.2%
517.8 515.1 −0.5%
41.1 40.7 −0.9%
19.9 19.6 −1.5%
27.9 24.9 −10.9%
89.0 85.2 −4.2%
The power matching is quite good, apart from for the pump power when the cooling is off in the early afternoon. The data indicates that the Cc pump is still on but that the flow is blocked by an internal valve in the TDC. This is a non-optimised operation of the real controller, and was thus not included in the simulation model. Additionally the TDC has the second water pump running at times when it is not modelled. However, these result in very small electrical use. The identification/calibration results are detailed in Tables 6 and 7. All simulated quantities are within 3% of the measured values apart from the operating time of the Dc pump, which is 4% longer on in reality. A final check of the system model was made by using the TRNSYS weather data for Borlänge [14] and picking out a similar period of weather data as for the calibration period. No inputs from the monitored data were used in the model. The performance figures for this simulation are compared with those from the monitored data in Table 8. This shows that there is a good agreement between the values, although the real boundary conditions were not exactly the same. The average ambient temperature for the simulation was 19.7 ◦ C compared to the monitored value of 20.9 ◦ C. This is the presumed reason for the fan power being significantly lower in the simulation compared to the monitored data.
5. Conclusions The operation during the cooling season of 2008 has revealed a number of weaknesses in the system design and operation. Although a number of weaknesses have been found, it should be stated that the TDC has worked reliably during the whole season. The main problem to be addressed is the high use of electricity to run the system. As it is now, the system provides a relatively low electrical COP and it is lower than that of the main compression chiller. The TDC system is capable of providing maximum thermal and electrical COP’s during the hottest period of around 0.50 and 4.6 respectively. However, the figures were only 0.41 and 2.1 for the complete monitoring period during the summer of 2008 respectively. The lower figures were due to continuous pump operation inside the TDC even during periods of no cold production as well as a period when no cold was produced due to an electrical fault in the overall controller. The figures for the complete system were 0.38 and 1.46 for the thermal and electrical COP respectively for the hot period and 0.29 and 1.0 for the whole summer. This shows significant heat losses between TDC and supply as well as high electricity use compared to cold production. The internal pumps in the TDC and high pump power in the connected circuits accounted for the majority of this energy. System simulation and parametric study [13] of this study determines how this can be reduced. There are a number of other causes of the relatively low thermal and electrical COP values of the entire system such as: • Heat exchanger in the driving circuit causes extra heat losses in the driving circuit of the TDC that significantly reduce the thermal COP of the system compared to that of the TDC. The heat exchanger results in lower operating temperatures as well. The ideal would be to have no extra heat exchanger.
• Heat exchanger in the chilled water delivery circuit causes heat gains in the delivery circuit both through normal gains through the insulation and components but also due to thermal energy from the two pumps used. In addition it forces the TDC to work at lower temperatures, reducing its cooling capacity. The ideal would be to have no extra heat exchanger. • The driving temperature available from the district heating network is lower than ideal for the TDC. It is on average 75–80 ◦ C, whereas 80–90 ◦ C would be more ideal. This affects the charging time of the TDC and thus also parasitic electrical use. • The operating times of the whole system are relatively short. The main compressor chiller is only operating on average 12–27% of the time for the months of the cooling season. This affects the economy as well as the electrical use of the TDC. • The TDC cannot deliver at full power with the normal operating conditions of the system. The system simulation model was calibrated in a three stage process against the monitored data. The final calibration stage showed very good agreement with the monitored data for a five day period with higher cooling loads in terms of both thermal and electrical energy quantities as well as running times of the pumps. The calibrated model is further used for parametric studies in order to find improved system design and control. The simulation study and parametric study of decentralised cooling in district heating network is described in [13]. Acknowledgments The authors wish to convey their sincere appreciation to European Commission (EC) for financial support of this project through FP6-2004-TREN-3 (Contract No. 019988), and similarly to the Swedish Energy Agency for funding via project: P22374-1. References [1] S. Lindmark, The role of absorption cooling for reaching sustainable energy systems, Licentiate Thesis, Division of Energy Processes, Royal Institute of Technology (KTH), Sweden, 2005. [2] M. Rydstrand, Heat-driven cooling in district energy systems, Licentiate Thesis, Division of Energy Processes, Royal Institute of Technology (KTH), Sweden, 2004. [3] L. Trygg, S. Amiri, European perspective on absorption cooling in a combined heat and power system—a case study of energy utility and industries in Sweden, Applied Energy 84 (2007) 1319–1337. [4] A. Egeskog, J. Hansson, G. Berndes, S. Werner, Co-generation of biofuels for transportation and heat for district heating systems-an assessment of the national possibilities in the EU, Energy Policy 37 (2009) 5260–5272. [5] PolySMART, EU-Integrated Project, Project No. 019988, FP6-2004-TREN-3. POLYgeneration with advanced Small and Medium scale thermally driven Airconditioning and Refrigeration Technology (PolySMART). www.polysmart.org. [6] ClimateWell AB, Product description – ClimateWellTM 10 (4th generation), Available at http://www.thermostat.com.tw/pdf/cognition 10.pdf (accessed 6.05.10). [7] M.R. Conde, Properties of aqueous solution of lithium and calcium chlorides: formulations for use in air conditioning equipment design, International Journal of Thermal Sciences 43 (2004) 367–382. [8] R. Olsson, M. Kaarebring-Olsson, S. Jonsson, A Chemical Heat Pump, Patent No: WO0037864, Sweden, 2000. [9] R. Boer, W.G. Haije, J.B.J. Veldhuis, Determination of structural, thermodynamic and phase properties in the Na2 S–H2 O system for application in a chemical heat pump, Journal of Thermochimica Acta 395 (2003) 3–19.
S. Udomsri et al. / Energy and Buildings 43 (2011) 3311–3321 [10] H. Ogura, T. Yamamoto, H. Kage, Efficiencies of CaO/H2 O/Ca(OH)2 chemical heat pump for heat storing and heating/cooling, Energy 28 (2003) 1479– 1493. [11] C. Bales, S. Nordlander, TCA Evaluation, Lab Measurements, Modelling and System Simulations. No. 3-9809656-4-3, Högskolan Dalarna, Borlänge, Sweden, 2005, ISSN 1401-7555. [12] F. Holmberg, Alternative Refrigerants for ClimateWell’s Sorption Cooling Technology, Master Thesis, Division of Energy Processes, Department of Chemical Technology, Royal Institute of Technology (KTH), Sweden, 2008.
3321
[13] S. Udomsri, C. Bales, A.R. Martin, V. Martin, Decentralised cooling in district heating network: system simulation and parametric study, Applied Energy, submitted for publication, Manuscript Number: APEN-D-10-01610. [14] TRNSYS, TRNSYS Coordinator, Solar Energy Laboratory, University of Wisconsin-Madison, 1500 Engineering Drive, 1303 Engineering Research Building Madison, WI 53706 – U.S.A. [15] C. Bales, O. Ayadi, Modelling of commercial absorption heat pump with integral storage, in: Proceedings Effstock 2009, CD-rom, Swedvac, Stockholm, Sweden, 2009.