Pergamon

www.elsevier.com / locate / solener

PII:

Solar Energy Vol. 69(Suppl.), Nos. 16, pp. 197208, 2000 2001 Elsevier Science Ltd S 0 0 3 8 0 9 2 X ( 0 0 ) 0 0 1 5 4 7 All rights reserved. Printed in Great Britain 0038-092X / 00 / $ - see front matter

INFLUENCE OF THE DHW LOAD PROFILE ON THE FRACTIONAL ENERGY SAVINGS: A CASE STUDY OF A SOLAR COMBI-SYSTEM WITH TRNSYS SIMULATIONS
ULRIKE JORDAN and KLAUS VAJEN
Marburg, FB Physik, D-35032 Marburg, Germany Universitat Received 20 June 2000; revised version accepted 8 November 2000 Communicated by ERICH HAHNE

AbstractIn this paper the inuence of domestic hot water (DHW) load proles with a constant total yearly heat demand is studied for a solar combisystem. Whereas so far simulation studies are usually done with a simplied DHW load prole (e.g. according to prEN 12977), a more realistic prole was generated on a 1-min time scale with statistical means. Assumptions about the distribution of the DHW consumption during the year, depending also on the weekday, and the time of the day were made. To study the inuence of the draw-off duration and ow rate as well as the daytime of DHW consumption, TRNSYS simulations were carried out with a variety of simplied and fairly realistic load proles. In this case study the investigated solar system consists of a storage tank with an internal thermosyphonally driven discharge unit. Despite a rather minor inuence of the draw-off ow rate on the fractional energy savings for the given system applying simplied proles: a fairly wide range of more than 2 percentage points of the fractional energy savings was found for the system with some constructive improvements or for applying more realistic DHW-proles. It can be concluded that the inuence of the DHW-load prole may not be disregarded, when combistores are compared. This is true especially for stores, for which duration and ow rate of a DHW draw-off have a severe inuence on the temperature stratication in the storage tank. 2001 Elsevier Science Ltd. All rights reserved.

1. INTRODUCTION

The fact that the temperature stratication in a store has a great impact on the solar energy gain of a solar system has been shown in many publications in the past. For example, Sharp and Loehrke (1979) showed that a relative performance advantage of systems with stratied over systems with well-mixed stores of 515% can be achieved. Therefore, much effort has been made to improve the thermal stratication in storage tanks. A lot of studies in the literature (e.g. Lavan and Thompson, 1977; Phillips and Dave, 1982; Morrison and Braun, 1985; Shyu et al., 1989) have analyzed the factors that inuence the stratication, like ow rates, temperature differences, geometry of inlets, stratiers, heat exchangers and the tank design. Andersen and Furbo (1999), for example, found a signicant decrease of the thermal performance caused by mixing during draw-offs for small solar domestic hot water (SDHW) systems. Consequently, the solar energy gain depends on

Author to whom correspondence should be addressed. Tel.: 1 49-6421-282-4148; fax: 1 49-6421-282-6535; e-mail: solar@physik.uni-marburg.de 197

the DHW-load prole in cases where the stratication of the storage tank depends on the DHWow rate, the duration of a draw-off, or the daytime of DHW-consumption. The impact of these quantities again depends on the construction of the store. In the recent years new types of combistores, for combined DHW and space heating (SH), have been developed. Some of these stores are equipped with an internal heat exchanger, thermally connecting the storage water, used for SH, and the DHW-cycle. Tests of these stores were carried out, for example by ITW (Stuttgart, Ger many) and SERC (Borlange, Sweden). It has been shown that the temperature stratication in the storage tanks of this kind may depend strongly on constructive details (Dahm et al., 1998, Druck and Hahne, 1998). Hampel et al., 1999, found in a simulation study a decrease of the collector output of almost 5% for a SDHW system of a multi-family house when the DHW draw-off is tapped in the evening instead of taking it in the morning. In contrast to that, so far simulation studies are usually carried out with a domestic hot water prole consisting of three draw-offs during the day with a constant ow rate of 10 l / min (prEN 12977).

198

U. Jordan and K. Vajen Table 1. Reference conditions for the load prole referred to a one family house with a mean consumption of 200 l / d and a load temperature of 458C. Four categories are dened with a mean ow rate, a constant duration of every DHW draw off, the medium number of draw offs during a day, and the standard deviation of the ow rates around the mean value. Furthermore the resulting mean values for the volume of one load and the volume during a day are shown Cat A: short l. Flow rate in l / min Duration in min Inc / day s Vol / load in l Vol / day in l Portion
a

In this case study, a solar combisystem with an internal thermosyphonally driven DHW heat exchanger is investigated and the dependence of the fractional energy savings on the ow rate due to daytime and draw-off duration is shown. For that purpose a DHW load prole was developed which should be more realistic than the commonly used DHW proles. It is described in the following section. Assumptions made are based on various studies about DHW consumption in Switzerland and Germany (Dittrich et al., 1972; Loose, 1991; Mack et al., 1998; Dichter, 1999; Nipkow, 1999; Real et al., 1999). In the third section, the model of the solar system that was implemented in the simulation tool TRNSYS (Klein et al., 1998), the reference conditions, and the applied TRNSYS deck are described briey. The mathematical model for the DHW discharge unit is shown in Section four. In Section ve, solar gains during the year of simulations applying the realistic DHW prole are compared to those applying the commonly used DHW prole. Furthermore, simple proles were used to analyze the inuence of the DHW ow rate, duration, and daytime on fractional energy savings. Storage temperature distributions, resulting from 2-day simulations, are shown. In the last Section the inuence of some variations of the realistic prole are discussed. The paper is related to Task 26 in the framework of the Solar Heating and Coding Programme of the IEA (International Energy Agency).
2. REALISTIC DHW LOAD PROFILE

Cat B: med. l. 6 1 12 2 6 72 36%

Cat C: bath 14 10 0.143 a 2 140 20 10%

Cat D: shower 8 5 2 2 40 80 40%

1 1 28 2 1 28 14%

Once a week.

2.1. Basic assumptions

Four categories were dened to describe the different types of loads: cat A: short load (washing hands, etc.) cat B: medium load (dish-washer, etc.) cat C: bath cat D: shower For each category a mean ow rate, a draw-off duration, the mean number of draw-offs during a day (inc / day), and standard deviation s were chosen, as shown in Table 1 for a daily mean draw-off volume of 200 l. The proles were generated separately for each category and superposed afterwards. The actual ow rate values in each minute are spread around the mean value with a Gaussian-distribution as shown in Fig. 2:

A load prole for the domestic hot water demand for a period of one year was generated. In order to take into account fairly realistic conditions, a time step of 1 min was chosen. The values of the ow rate and the times of the draw-offs were selected by statistical means. The rst 3 days of the prole with a daily mean draw-off volume of 200 l are shown in Fig. 1.

~ 2V ~mean )2 2 (V 1 ~ ) 5 ]]] ]]]]] prob( V exp (2p)0.5 s 2s 2

(1)

With this approach it is assumed that there is no correlation between the weather data and the

Fig. 1. Realistic load prole, Jan. 1 to Jan. 3, generated with statistical means. Mean draw-off volume of the year: 200 l / d.

Fig. 2. Realistic load prole: Total duration of draw-offs during one year as a function of the DHW ow rate. These are distributed with a gaussian function (Eq. 1) around the mean values, also given in Table 1.

Inuence of the DHW load prole on the fractional energy savings

199

DHW load prole. A probability function, describing variations of the load prole during the year (also taking into account daylight saving time), the weekday, and the day was dened for every category. The course of probabilities during the year is described by the product of probability distributions during the year, during the day, at weekdays and during the holiday season:
probx ( year) 5 probx (season)*probx (week )*probx (day)*probx (holiday) with x : category A, B, C, D Fig. 4. Distribution of probability for category C: Bath tub lling (white columns). In contrast to category C, no day of the week was privileged in categories A, B, and D.

prob x (season): The course of probabilities during the year is described by a sine-function with an amplitude of 10% of the daily average discharge volume (Fig. 3). Mack et al., 1998, found variations of the energy consumptions according to a sine function with an amplitude of 25% and a maximum during winter and a minimum during summer time. This variation is due to variations of the cold water temperature, variable consumption during the year, and due to holidays, taking place mainly in the summer. Since holidays are taken into account separately (see below) and the cold water temperature is not dened by this load prole, the amplitude given by Mack et al. was reduced. prob x (week ): At different days of the week the probability for taking a bath and the mean distribution for the total volume per day are shown in Fig. 4. For the categories A, B, and D the probability for every day of the week is assumed to be the same. The average of DHWconsumption for the four categories is according to the results found by Dichter, 1999. prob x (day): The assumptions for the daily distribution used are shown in Fig. 5. For a short and medium load the probability is distributed equally between 5:00 and 23:00 h. For the shower bath, a clear peak in the

morning and for the bath tub lling (which mainly occurs on weekends), a peak in the evening were applied. prob x (holiday): For every 100 l / d of consumption the probability was set to zero during a period of 2 weeks during the summer. Therefore, for a one family house with a total daily mean consumption of 200 l / d, two periods in which the consumption is reduced by one half are taken into account. The start day of every holiday period is delivered by a random generator. The generator is set in a way that for a consumption of 200 l / d the two periods of reduced consumption do not coincide, but start on Jul. 14 and Aug. 8, respectively. The function prob x (holiday) is dened as follows:
mean volume of daily load 2 reduced volume probx (holiday) 5 ]]]]]]]]] mean volume of daily load

hence with a consumption of 200 l / d:

1 prob(holiday) 5 ] 2 prob(holiday) 5 1

Jul. 14..27, Aug. 8..21 else

2.2. Method
The cumulated frequency method was used to distribute the draw-off incidences among the year according to the probability function. As shown in Fig. 6, the probability function prob x ( year) was

Fig. 3. Realistic load prole: Daily DHW-load volume in the course of the year. Mean value: 200 l / d, bold line: prob x (season).

Fig. 5. Distribution of prob x (day) during the day.

200

U. Jordan and K. Vajen

Fig. 6. Accumulated frequency method: Integral of the probability function prob x ( year).

integrated over the year and normalized. Also the number of draw-offs during the year was calculated. The same number of values between zero and one were generated randomly and assigned to a ow rate in the order of occurrence. With Fig. 6, every random value (value on the y -axis) was then assigned to a minute of the year (value on the x -axis). In this way load proles were generated for different demands as well. The basic load is 100 l / day. Proles were generated in dual order (100, 200, 400, 800 l), with different initial random values . Therefore, it is possible to get any load prole for a multi-family house with consumptions in steps of 100 l / day by superposition of the generated les. For different consumptions the number of incidences, shown in Table 1, are adjusted to the mean draw-off volumes.

3. INVESTIGATED SOLAR SYSTEM

3.1. TRNSYS deck and reference conditions

A solar combisystem for a one family house in Zurich is regarded, with fractional energy savings

of about 25%. A scheme of the TRNSYS deck used is shown in Fig. 7. The weather as well as a set of space heating data were implemented as data les. The space heating data from IEA-Task 26 were used with input and output temperatures and the ow rate for the space heating cycle (Streicher and Heimrath, 2000). The DHW-load was either taken from a data le in case of using a statistically generated prole. Alternatively, conventional load patterns were implemented with the Type 14 time dependent forcing function. For the thermosyphonally driven discharge unit of the combitank a user contributed plug-ow heat exchanger model was developed (Type 104). Other than that, standard components, including the Type 140 multiport storage type (Druck, 1994), and a matched ow collector model (Isakson, 1995) were used. The solar system consists of a 14 m 2 at plate collector, an external solar heat exchanger, and an 817 l combistore with an internal thermosyphonally driven DHW heat exchanger. A scheme of the combitank is shown in Fig. 8. Some of the reference conditions assumed, are listed in Table 2. When domestic water is heated, the water inside an open box surrounding the heat exchanger is cooled, and ows through an adjacent tube to the bottom of the store. Depending on the storage temperatures and on the ow rate, the storage water ow needs to be slowed down, in order to cool it in the heat exchanger sufciently. This is done by a valve in the vertical tube, placed below the heat exchanger (Fig. 9). The valve is connected to a thin cylinder containing an expansible liquid that is placed at the domestic water outlet. For stationary conditions, the angle of the valve is a (linear)

Fig. 7. Scheme of the TRNSYS-deck with a new developed user contributed component Type 104 for the discharge unit (plug ow heat exchanger model combined with thermosyphon driving pressures). The numbers indicate the TRNSYS Type numbers used.

The proles are available from the authors on request.

Inuence of the DHW load prole on the fractional energy savings

201

Fig. 8. Combistore with thermosyphonally driven discharge unit: During a DHW draw-off storage water is driven inside the tank by density differences. Table 2. Reference conditions Auxiliary set temp. Collector area Collector orientation Coll. specic ow rate Daily DHW-load volume Design temp. SH-distr. syst. DHW-set temperature Domestic cold water temp. Flow rate auxiliary heating Location Space heating demand Store volume Weather data (57 1 5)8C 14 m 2 south, tilt angle: 458 40 l / m 2 h 200 l 408C / 358C (45-1)8C (9.766.3)8C, sine-func., min at March 1. 430 kg / h Zurich 8400 kWh / a (140 m 2 , 60 kWh / m 2 a) 817 l Meteonorm, Dt 5 1 h

function of the domestic water outlet temperature. Therefore the pressure drop at the valve increases, if the DHW-temperature rises. This leads to a decrease of the storage water ow rate and the primary outlet temperature of the heat exchanger. In this way, it can be avoided that hot water ows from the top to the bottom part of the store.

the electric energy demand for pumps was taken into account:
hburner Q burner,sol 1 Q pump,sol ? ]] hpump fsave 5 1 2 ]]]]]]]]]] hburner Q SH 1 Q DHW 1 Q store.ls,conv 1 Q pump,conv ? ]] hpump

(2)

with Q store.ls,conv 5 644 kWh,

3.2. Denition of the fractional energy savings

According to prEN 12977 the energy savings shall be calculated by comparing the gross auxiliary energy demand of the auxiliary heater of the solar heating system to the gross auxiliary energy demand of a conventional heating system. Additionally to the denition given in the norm,

hburner 85% ]] 5 ]] 5 2.125, hpump 40%

Ppump 5 30 W, Pburner 5 10 kW. The value of Q store.ls,con v was calculated with values of UA store , a storage size, and set temperatures according to prEN 12977. To calculate the running time of the pumps to load the store of the conventional system, the total energy demand of the conventional system (including storage losses) was divided by the assumed power of the burner. Running times of the solar system pumps result from simulations.
4. MODELLING

Fig. 9. (a) Regulation device, composed of a valve placed in a vertical tube below the heat exchanger and a thin cylinder containing an expansible material. (b) angle of the valve w. (c) Mixer for the domestic water. Validation of the heat exchanger model (without store).

4.1. Relations of the thermosyphon loop

The model of the discharge unit is described in

202

U. Jordan and K. Vajen

detail in Jordan et al. (1999). In the following, a short formulation of the equations that are implemented in Type 104 is given. A common approach to model a density driven circulation loop is the one dimensional steady state momentum equation for incompressible ow, as the balance of the pressure and frictional forces. The driving forces for the storage water ow in the discharge unit are expressed by density differences. Frictional forces are described by the kinetic pressure drop of the storage water. The pressure drop coefcient is a function of the geometry of the discharge unit, the angle of the valve, and the ow rate of the storage water. h tube n g ]] (r 2 ri ,tube ) n i 5 1 i ,stor
m h hx 1 ]] (r 2 ri ,hx1 ) m 2 n i 5 n 1 1 i ,stor

F O

G
(3)
Fig. 10. (a) Comparison of measured (black) and modeled (grey) outlet temperatures of the heat exchanger. (b) Measured value of DHW ow rate and calculated storage water ow rate (by the energy balance).

2 2 prim,out v prim,out store v stor 5 0.5 ? ( 2 stor r ) tube r

~stor ,d, . . . ), and r,n, with vprim,out 4 vstor , 5 (w,V and : density, velocity, and pressure drop coefcient, respectively. The index i indicates the temperature nodes in the store, in the heat exchanger (hx) and in the vertical tube below the heat exchanger. Due to the fact that the velocity of the storage water in the vertical tube is much higher than the upwards velocity of the storage water in the tank, the second term on the right hand side in Eq. (3) may be neglected. The transient behaviour of the valve is described by ~ 1 w 5 a (T dom,out 2 T b ) tw
(4)

4.2. Experimental validation

Temperatures on a vertical ledge at 15 positions in the storage tank as well as inlet and outlet temperatures of the heat exchanger and the domestic water ow rate were measured with thermocouples. The storage temperatures at the ledge were input data to the simulation. In Fig. 10 some measured, modeled, and calculated time series are shown. The domestic water ow rate was varied during the measurement. The reaction of storage water to the domestic water ow changes occurs smooth and with a certain time delay. The measured and calculated values of the tube outlet temperature differ considerably only for low storage water ow rates. These effects, however, do not inuence the energy balance severely. As shown in Fig. 10, the domestic water outlet temperature is described quite well by the model.

The value for t was determined empirically by the response of the angle of the valve to step changes of T dom,out . The heat exchanger is modeled with the energy equation (Eq. 5a and 5b). A plug ow heat exchanger model was implemented into Type 104.
T dom T dom ~ dom c p ? ]] Cdom ]] 5 UA ? (T dom 2 T stor ) 2 M t x *

(5a) T stor T stor ~ stor c p ? ]] Cstor ]] 5 UA ? (T dom 2 T stor ) 1 M t x * (5b) x * is the normalized length of the heat exchanger. The heat capacity rate (UA) was described as a function of the domestic water ow rate.
5. ONE YEAR SIMULATION RESULTS

In prEN 12977 reference conditions necessary for simulation studies are proposed. According to this norm the DHW prole is supposed to be composed of three draw-offs during the day at 7 a.m., 12 a.m., and at 5 p.m. The total load should

Inuence of the DHW load prole on the fractional energy savings

203

be divided up in the proportion 2 / 5, 1 / 5, 2 / 5, respectively. This prole, with the time of the evening drawoff changed to 7 p.m., will be considered in the following as the conventional prole. None of the demand distributions described in Section two are taken into account for the conventional prole. The only time dependent variable used is the domestic cold water temperature, which is described by a sine-function. Since the same yearly total DHW volume is assumed for both, the realistic and the conventional prole, the energy demand for the two proles differ slightly. For the simulations shown in the following and with the cold water temperature as described in Table 2, the energy demand is reduced by about 13.5 kWh ( , 0.5% of the overall DHW and about 0.1% of the overall heat demand), when applying the realistic compared to the conventional load prole. Therefore the different energy demand applied do not have a dominating inuence on the simulation results shown in the following. In Fig. 11 monthly mean values of the collector energy gain Q hx reduced by the store losses Q store,ls are compared. The following results are indicated:

April, is about 3.4 times the December value for the realistic prole. During the months November to January 72% of the solar gain are obtained compared to the values between June and August. These values depend strongly on the cold water temperature distribution during the year, as well as on the charging strategy of the store and the layout of the conventional heating system.

5.2. Differences of the solar energy gain due to the DHW proles
The values of the solar energy gain for the realistic prole are much lower during the summer than the values applying the conventional prole. This is mainly due to the reduced consumption during the summer (sine-distribution of the probability function) and the holiday season that is taken into account with the realistic prole.

5.3. Inuence of the ow rate and draw-off time of the day

In order to study the inuence of the DHW ow rate on the fractional energy savings, three types of DHW load proles similar to the conventional one were created with a total load of 200 l / d: (a) the load-division into three parts of 2 / 5, 1 / 5, 2 / 5 was kept (3 3 daily), (b) the total load was set starting at 7 a.m. (morning), (c) starting at 12 a.m. (noon). The ow rate and duration of draw-offs were varied as shown in Table 3. For example, for prole (a) the lowest ow rate was 4 l / min with the durations 20 min at 7 a.m., 10 min at 12 a.m, and 20 min at 7 p.m.

5.1. Distribution of the solar energy gain during the year

The highest solar gains are not obtained during the summer, but during spring and fall. This is mainly due to the distribution of the space heating demand during the year. The seasonal variations of the solar energy gains in the course of the year are quite small compared to the ones for typical SDHW-systems. The maximum value, obtained in

Fig. 11. Monthly mean values of the heat transfered in the heat exchanger of the collector circuit reduced by the storage losses for applying the conventional and realistic prole (left y -axis). Also the total heat demand as the sum of Q DHW and Q SH is shown (right y -axis).

204 Table 3. Flow rates and durations DHW draw-offs Flow rate (l / min) (b), (c) Total duration (min) 50 40 30 20 15 10

U. Jordan and K. Vajen

(a) Distribution morningnoonevening (min) 201020 16816 12612 848 636 424

between the values of fsa v e is about 3.5%-points for the system without using a ow regulation device.

4.00 5.00 6.67 10.00 13.33 20.00

5.4. Inuence of ow rate and duration

With operation of the ow regulation device the fractional energy savings, fsa v e , are almost independent of the DHW ow rate. However, if the regulation is not in operation, fsa v e depends on the ow rate considerably: Applying three daily draw-offs as described in Table 3, the maximum difference between the values of fsa v e is more than 2%-points. Related to the value of fsa v e with a ow rate of 10 l / min (as proposed in the prEN) fsa v e varies by about 61%-point. These differences of fsa v e for simulations without a ow regulation device can be explained by

One year simulation results are shown in Figs. 12 and 13. In Fig. 12 the curves at the top show results with the storage water ow regulation in operation, the bottom curves show simulation results for a system without ow regulation. The gure indicates that the maximum difference

Fig. 12. Fractional energy savings. One year simulations with one or three draw-offs during the day and with the realistic load prole described in Section two. For the simplied proles the DHW ow rate was varied in the range between 4 and 20 l / min according to Table 3.

Fig. 13. Fractional energy savings. One year simulations with one draw-off during the day. The draw-off time was varied between 7 a.m. and 7 p.m.

Inuence of the DHW load prole on the fractional energy savings

205

the relation of the storage to the domestic-water ow rate

~stor V r []] ~DHW V

The value of r turns out to be higher for low than for high domestic-water ow rates during a drawoff for equal reference conditions. This is due to the fact that the pressure drop of the storage water ow depends on the storage water ow rate, as shown in Eq. (3). For this reason, a high storage water ow is slowed down to a higher extent than a small one. The higher the value of r, the higher the storage water outlet temperature T store,out of the heat exchanger and the more heat is delivered from the top to the bottom of the tank. For example, short term simulations with one morning draw-off and durations as listed in Table 3 showed: with ow regulation:

depend strongly on the temperatures in the storage tank. Nevertheless they indicate the tendency that the value of the DHW-ow rate does not play an important role for the temperature stratication, if the system is designed with the ow regulation device. Without ow regulation, the design of the discharge unit is more suitable for high DHWow rates, whereas the pressure drop should be enhanced for small DHW-ow rates.

~DHW 5 20 l / min, T store,bottom 5 208C, (r 1 for V

T store,aux 5 578C).

5.5. Draw-off time of the day

fsa v e depends on the time of the draw-off for both constructions. The values resulting from the noon prole are distinctly above the other values. With ow regulation, the values of fsa v e for the morning and an evening load (not shown in Fig. 13) are about the same. If the DHW-load is taken at noon, the values of fsa v e are about 0.7%-points higher than the values for the loads taken in the morning. Without ow regulation, the mean value of fsa v e resulting from the noon proles differs from the one from the morning proles by about 1.1%-points. In Fig. 14 simulation results with one daily draw-off at different times of the day are shown.

~DHW 5 4 l/min and for V ~DHW 5 20 l/ min; 1 for V r

without ow regulation:

~DHW 5 4 l / min and 3.6 for V r ~DHW 5 20 l / min. 1.5 for V r

do not hold generally, but These mean values r

Fig. 14. (ad) Two-day simulations with morning or evening draw-offs. Ten temperatures in the storage tank are shown at the normalized heights: 0.05, 0.15,.. 0.95. Initial storage temperature: 208C. During the rst hour, the three top layers are heated by auxiliary energy. DHW ow rate: 10 l / min; draw-off volume per day: 200 l / d. (a) and (c) Simulations with ow regulation, (b) and (d) without ow regulation.

206

U. Jordan and K. Vajen

For all of the proles shown, there is a maximum value of fsa v e around the early afternoon.

5.6. Comparison of simplied and realistic DHW proles and the inuence of ow regulation
As shown in Fig. 12, the values of the fractional energy savings for applying the realistic compared to those applying the simplied proles for the same system design are reduced by up to more than 3%-points. This means, regarding absolute values, that the collector energy transfered in the heat exchanger reduced by the store losses varies by about 430 kWh / a due to different (simplied) DHW-proles. Without ow regulation the values applying the realistic prole are close to the lowest ones applying the simplied proles. The differences between the results with and without ow control are about 2.3%-points for the realistic prole. The mean DHW ow rate of the realistic prole is 6.9 l / min. The values of fsa v e for the realistic prole are lower than the values for the simplied proles with morning draw-offs and a DHW ow rate of about 6.9 l / min. This is true, although an even distribution of draw-offs during the day has a positive impact on the fractional energy savings. The reason for the lower values of fsa v e for the realistic prole is that the mean number of draw-offs per day is about 42 compared to one or three draw-offs per day for the simplied proles and that the thermosyphonally driven storage water ow does not stop instantaneously after a draw-off. Therefore a certain amount of warm water ows from the top to the bottom of the tank after each tap.

in the store in an equal vertical distance are shown. The calculations were done with 100 temperature nodes, with the one dimensional multiport storage TRNSYS-Type 140. The whole store is heated up in the rst day until about 4 p.m. to a temperature of 608C. Therefore, at the rst day more auxiliary energy is needed in the simulations shown in Fig. 14a and 14b (morning), and some more solar energy gets into the store than for the evening draw-offs. At the end of the day, the store is hot in the cases of the morning draw-offs. For the evening draw-offs, the lowest temperature in the store is about 258C with ow regulation, and it drops to about 408C without ow regulation. The temperature at the bottom of the store is below 308C during the night with ow regulation (Fig. 14c) compared to above 408C, without (Fig. 14d). Therefore, during the night the store losses are reduced for the evening compared to the morning draw-offs. At the second day again the bottom part of the store is heated up to 608C at 4 p.m. Until then, there is more solar energy gain for the morning than for the evening proles. In this example, for morning draw-offs more solar energy is delivered into the store, however the overall heat losses are higher than for evening draw-offs. Therefore in this case the solar system gain ( Q hx 2 Q store,ls ) turns out to be almost exactly the same for the morning as for the evening draw-offs.

5.8. Variations of the realistic load prole

The inuence of assumptions made for the distributions of the DHW-consumption for the realistic load prole, described in Section 2 was investigated. In Fig. 15 1-year simulation results

5.7. Stratication in the storage tank: 2 -day simulations

The following considerations will not give nal explanations for the results shown in Section 5.3, but illustrate some tendencies that are induced by different draw-off patterns and by the implementation of a ow regulation device. For this kind of investigations, reference conditions play an important role. As an example, extreme reference conditions, a cold storage tank and fairly high insulation, were chosen. In Fig. 14ad the temperature stratication in the storage tank is shown for 2-day simulations for a warm weather period in spring. The initial temperatures in the store were 208C. The auxiliary heater delivers a set temperature of 628C when the temperature at a sensor, placed at a normalized height of 0.8, sinks below 578C. Ten temperatures

Fig. 15. Difference between the values of fsa v e applying the prole as described in Section two and a slightly changed prole (each with a mean DHW draw-off of 200 l / d and with ow regulation). In simulations (a)(d), distributions of the applied DHW prole were neglected one after another (week, day, . . . ), in (e) none of the distributions was taken into account.

Inuence of the DHW load prole on the fractional energy savings

207

are shown for realistic load proles without taking into account the distributions for the DHWconsumption (a) during the weekday (see Fig. 3) (b) during the day (see Fig. 4) (c) during the year (sine function with minimum in the summer time) (d) during the holiday season (e) of any of the distributions ad. For simulation b and e it was assumed, that there was no DHW-consumption between 11 p.m. and 5 a.m. During the rest of the day the probability for DHW-consumption was assumed to be equally distributed. Each distribution had a negative impact on the fractional energy savings. When none of them was applied, the fractional energy savings increased by 1.1%-points (simulation e) compared to the prole described in Section two. According to the results shown in Fig. 15, taking into account the holiday season had the highest inuence (0.4%-points), followed by the yearly sine-distribution (0.3%points), and the daily distribution (0.1%-points), respectively. Almost no difference was found neglecting the distribution for different weekdays.

4. Variations of the realistic DHW prole concerning distributions of the DHW-consumption showed, that the highest inuence was obtained by taking into account a holiday period during the summer, followed by a yearly distribution of the DHW consumption described by a sine-function, and by the daily distribution. The inuence of taking into account the different probabilities at different weekdays could be neglected. It can be concluded that the inuence of the DHW load prole may not be disregarded, for a comparison as well as for an optimization of combistores. This is especially true if durations and ow rates of the DHW draw-offs have a severe inuence on the temperature stratication in the storage tank. An optimization for only one ow rate and a small number of draw-offs may lead to non optimal solutions for realistic reference conditions.

6.1. Final remark

Compared to the paper presented by the same authors at the EuroSun Conference 2000, Copenhagen, the evaluations in this paper are carried out with slightly different reference conditions, e.g. a different seasonal variation of the cold water temperature. Furthermore, corrections to the results of the fractional energy savings for the realistic proles were made in this paper.

6. CONCLUSIONS

A realistic DHW load prole with a 1 min time step was generated with statistical means. Inuences of the DHW ow rate, the draw-off time of the day, for applying realistic and simplied DHW proles, and for two different designs of the discharge unit on the fractional energy savings fsa v e were investigated. 1. A comparison between simplied and more realistic load proles shows differences in the fractional energy savings for the investigated system designs of up to more than 3%-points. During the summer less solar energy is delivered applying the realistic DHW prole. 2. If the discharge unit is not designed properly, fsa v e varies by about 2%-points only depending on the DHW ow rate if one or three daily draw-offs are applied, similar to the DHW load pattern proposal by prEN 12977. For a system with a proper ow regulation device, no marked dependence on the ow rate was found. 3. Varying the draw-off time of the day for simplied DHW load patterns with one or three draw-offs per day, the highest values of fsa v e were found for draw-offs in the early afternoon. The differences of fsa v e for morning and noon proles differ by about 1%-point.

NOMENCLATURE
cp C d fsa v e g h ~ M P prob Q T UA ~ V x* Greek a r w n t h specic heat capacity, J / kgK heat capacity, J / K diameter, m fractional energy savings gravity constant, m / s 2 height, m mass ow rate, kg / s power (of burner and el. pump) / W probability energy (heat) / kWh temperature, K heat transfer coefcient*area, W/ K volume ow rate, m 3 / s normalized length of the heat exchanger

transfer coefcient, 1 / K density, kg / m 3 angle of the valve, 8 velocity, m / s time constant of the reaction of the valve, s pressure drop coefcient efciency

208 Indices aux b conv DHW dom hx1 i m n out prim,out sol stor store,ls SH

U. Jordan and K. Vajen H. and Hahne E. (1998). Test and comparison of hot Druck water stores for solar combisystems. In: Proceedings of EuroSun ISES Europe Solar Congress, Goetzberger A. (Eds), Vol. 2, III.3.3, Portoroz, Slovenia. Hampel M., Uecker M., Krause M., Vajen K. and Ackermann H. (1999). Einu des verbrauchsprols auf den solaren Ertrag ausgewahlter groerer Solaranlagen. In: Proceedings of 9. Symposium Thermische Solarenergie, (Eds), pp. 375379, Staffelstein, Germany. Isakson P. (1995). Solar collector model for testing and simulations, ISSN 0284-141X, Stockholm, Sweden. Jordan U., Vajen K., Knopf B., Spieler A. and Hilmer F. (1999). Modelling of a thermosyphonally driven discharge unit of a storage tank. In: Proceedings of ISES Solar World Congress, Jerusalem, Israel. In press. Klein S. A. et al. (1998). TRNSYS 14.2 Users Manual. University of Wisconsin, Solar Energy Laboratory. Lavan Z. and Thompson J. (1977) Experimental study of thermally stratied hot water storage tanks. Solar Energy 19, 519524. Loose P. (1991). Der Tagesgang des Trink-WarmwasserBedarfes, HLH42, Nr. 2. Mack M., Schwenk C. and Kohler S. (1998). Kollektoranlagen im Geschowohnungsbau eine Zwischenbilanz. In Proceedings of 11. Internationales Sonnenforum, (Eds), pp. Germany. 4552, Koln, Morrison G. L. and Braun J. E. (1985) System modeling and operation characteristics of thermosyphon solar water heaters. Solar Energy 34, 389405. Nipkow J. (1999). Warmwasser-Zapfungsverhalten. Schlus sbericht. Industrielle Betriebe der Stadt Zurich, Zurich. Phillips W. F. and Dave R. N. (1982) Effects of stratication on the performance of liquid-based solar heating systems. Solar Energy 29, 111120. Real M., Nipkow J., Tanner L., Stadelmann B. and Dinkel F. (1999). Simulation Warmwassersysteme. Schlussbericht Forschungsprogramm Wasser, Eidgenossische Drucksachenund Materialzentrale, Bern. prEN 12977 (1997). Preliminary European Standard. Thermal solar systems and components: Custom built systems. Test methods. Sharp M. K. and Loehrke R. I. (1979) Stratied thermal storage in residential solar energy applications. J. Energy 3, 106113. Shyu R. -J., Lin J. Y. and Fang L. J. (1989) Thermal analysis of stratied storage tanks. Transactions of the ASME 111, 5461. Streicher W. and Heimrath R. (2000). Reference conditions for solar combisystems: space heat demand. In: Minutes of the 4. Experts Meeting, IEA SHC-Task 26: Solar Com bisystems, Borlange, Sweden.

auxiliary energy of the solar system starting point of the regulator temperature interval of the valve conventional system domestic hot water (system) domestic water heat exchanger, primary side nr. of node, describing temperature layer max. nr. of nodes in the vertical tube max. nr. of nodes in the heat exchanger outlet (temperature) primary outlet temperature (heat exchanger) solar system storage water in heat exchanger storage heat losses space heating water (system)

Acknowledgements The authors would like to express their Stuttgart, Gersincere thanks to Harald Druck, Universitat many, Thomas Schabbach, Wagner & Co. Solartechnik, Ger Marburg, Germany, and to Prof. Ackermann, Universitat many, for their valuable assistance of this research study. Also the authors would like to thank the company Wagner & Co. Solartechnik for their nancial contribution to this work.

REFERENCES
Andersen E. and Furbo S. (1999). Thermal destratication in small standard solar tanks due to mixing during tapping. In: Proceedings of ISES Solar World Congress, in press. Dahm J., Bales Ch. and Lorenz K. (1998) Evaluation of storage congurations with internal heat exchangers. Solar Energy 62, 407417. Dichter E. (1999). Dusch- und Badeverhalten. Bericht zu einer Reprasentativumfrage, Eidgenossische Drucksachen- und Materialzentrale, Bern. Dittrich A., Linneberger B., Wegener W. (1972). Theorien zur Bedarfsermittlung und Verfahren zur Leistungskennzeich nung von Brauchwasser-Erwarmern. HLH23, Nr. 2. Druck H. (1994). Weiterentwicklung und Validierung des solare Warmwasserspeicher 4Port fur das Modells fur Simulationsprogramm TRNSYS, Diploma-Thesis, Univer Stuttgart. sitat