INTERMODEL COMPARISON AND EXPERIMENTAL
Contents
1. 3 rd 220L day 80 EXP TYPE 534 40 amp 70 TYPE 4 TYPE 340 2 TYPE 60 Qaa TrPE38 e 135 z 3 l x v 3 Ss O 2 z 3 Time hour Figure 9 Lower thermostat temperature as a function of time for a water draw of 450 and 220 L day 693 Proceedings of Building Simulation 2011 12th Conference of International Building Performance Simulation Association Sydney 14 16 November This is a limitation of the modelling approach where entering water is supply by the bottom of the tank and large plugs of fluid can be aggregated reducing the accuracy of the modelled thermostat temperature Table 3 presents the total energy use for the last day in a 3 day simulation The standby test result T 0 only requires electrical heating to compensate for standby losses The model simplifications for de stratification TYPE 4 and nodal segmentation TYPE 38 result in a larger error All nodal models with de stratification underestimate the thermal losses by approximately 10 which is related to the global tank loss coefficient Uank approximation This error could have been removed by using a tuned value for the tank global loss coefficient based on the standby test For water draw tests T gt 0 all models give similar results with a small increase 6 compared to the standby test in the underestimation of the overall energy used Again TYPE 60 results are af2. 2011 12th Conference of International Building Performance Simulation Association Sydney 14 16 November temperature difference Uj yp described in SRCC TM 1 2008 This U value is calculated as MaC 7 Tank inia T U immo Ta a Ank FUME decay env 2 Tan final Tn The results of the standby test gave a Ug joss of 1 05 W m K with a measured steady state heat loss rate of 93 W a total tank surface of 2 6 m and an average temperature difference Tinean tank Tmean ambiant Of 34 C The value obtained for Urmrp is virtually the same 1 04 W m K The value of 1 05 W mK was used in all models Studies e g Cruickshank et al 2010 have shown that assuming an average U value based on these methods can lead to errors up to 10 in predicting the storage thermal losses Even if better results can be obtained by calibrating the model with experimental data the previous U value results will be considered sufficient in a context where the objective is to obtain a good representation of the entire stock of water heaters Heat diffusion rate De stratification in the storage tank is mainly driven by thermal conduction within the fluid and conduction along the tank wall Therefore the effective conduction kep in the tank is represented by the sum of Kgyig 0 63 W mK for water and Ak This last term may be calculated as Klein et al 2010 A tank_wall tank wall A 3 c water Ak k Where kank wa
3. CONCLUSIONS Five one dimensional models of electric water heaters have been investigated and compared to experimental data The main objective of this study was to assess the usefulness of the different models for demand side management studies The results have shown that for a typical electric hot water heater found in Qu bec and a typical hot water draw models based on a nodal approach give a better performance than plug flow models The tested plug flow model TRNSYS standard TYPE 38 uses a variable number of segments that depends on operating conditions but this number is never large enough to reproduce the stratification observed experimentally TYPE 38 is therefore not recommended Heat diffusion in the tank has a significant impact on energy performance results especially for small tank turnover ratios Therefore models that do not take in consideration de stratification such as TRNSYS standard TYPE 4 are not recommended None of the tested models were able to correctly represent the vertical temperature profile in the mixing zone but the models that do consider de stratification offer a better approximation TYPE 60 exhibited numerical errors that affected the results significantly in some of the tests and is not recommended Non standard TYPE 340 is reasonably accurate and computationally very efficient compared to other nodal models However the limitation to one heating element could be a barrier to more general us
4. Proceedings of Building Simulation 2011 12th Conference of International Building Performance Simulation Association Sydney 14 16 November INTERMODEL COMPARISON AND EXPERIMENTAL VALIDATION OF ELECTRICAL WATER HEATER MODELS IN TRNSYS Yannick Allard Micha l Kummert Michel Bernier and Alain Moreau cole Polytechnique de Montr al D partement de g nie m canique P O Box 6079 succursale centre ville Montr al Qu bec H3C 3A7 Canada Laboratoire des Technologies de l nergie LTE Hydro Qu bec P O Box 990 Shawinigan Qu bec GON 7N5 Canada ABSTRACT This study compares the performance of five electrical water heater models in the TRNSYS environment The main capabilities modeling assumptions and performance of the models are first assessed using an intermodel comparison The main criteria for comparison are domestic hot water supply temperature power demand time dependent profile and overall energy use and vertical temperature distribution in the tank Experimental data are used to validate each model for one specific type of water heater and a selected water draw profile Finally the paper makes recommendations for selecting a model and configuring the model parameters in order to minimize the impact of modeling simplifications INTRODUCTION In the province of Qu bec Canada more than 90 of homes are served by electric water heaters with a total of over 2 7x10 units Domestic water heati
5. Temperature in Tank C 25 30 35 Temperature in Tank C Figure 4 Tank vertical temperature distributions during a water draw If the cold water inlet is not located at the bottom of the tank a phenomenon known as temperature inversion can occur during a water draw event In this case the temperature at the cold water inlet is lower than the temperature below This situation cannot physically persist in a real tank and all models have a way to cope with these transient situations In general the approach is to completely and 69 Proceedings of Building Simulation 2011 12th Conference of International Building Performance Simulation Association Sydney 14 16 November instantaneously mix the colder node with the node immediately below and repeat the process until no inversion remains Some models TYPE 534 and 340 allow some control on the speed at which thermal inversion is removed by using a non infinite mixing flow rate between inverted nodes Fully mixed TYPE534 Omin 120 0 5 min 1 min 100 1 5 min f 2min Position in Tank cm o Mixing flow rate TYPE 340 Omin 120 0 5 min 1 min 100 1 5 min 2min Position in Tank cm o Temperature in Tank C Figure 5 Influence of temperature inversion numerical routine on tank vertical temperature distributions during a single water draw event The effect of instantaneous mixing and controlled mixing flow r
6. ate approaches are compared in Figure 5 for a water draw event More specifically TYPE 534 with an instantaneous mixing flow routine and TYPE 340 with default mixing flow rate value have been used during this 1 5 minute withdrawal The Figure shows that temperature inversion is only significant during the time step when cold water enters the tank With this relatively small water demand no difference in the vertical temperature distribution is noticeable after the water draw event Electric heaters charging phase This comparison considered the period just after 10 AM when the electric heater is re activated after a peak shaving episode Again the last day in a 3 day simulation is used Simulation results presented in Figure 6 show the evolution of the tank vertical temperature distribution during the charging phase Similar to the water draw profile test de stratification has an impact on the temperature profile at the bottom of the tank Part of the heat released by the electric element is transferred by conduction to the lower nodes As mentioned before TYPE 4 ignores this effect which has an impact on the temperature distribution at 10 AM and also introduces a time lag in the profile evolution TYPE 340 again shows the impact of using a controlled flow rate to suppress temperature inversions in Figure 6 node containing the heating element is warmer than the nodes above it This detail has been experimentally validated Atabak
7. derestimate temperature for nodes located under the heating element T to Tyo The vertical temperature distribution obtained for the mixing zone right after a peak shaving episode should be interpreted with care 60 T 50 F 40H sy L j 30 20 y EXP TYPE 534 TYPE 4 TYPE 340 TYPE 60 10 60 50 S o p 20 J EXxP TYPE 534 TYPE 4 TYPE 340 TYPE 60 10 60 50 oO S 2 20 EXP TYPE 534 TYPE 4 TYPE 340 TYPE 60 10 48 54 60 66 72 Time hour Figure 10 Temperature for different height in the mixing zone vs time for a water draw of 450 L day No model is able to represent correctly the height of the mixing zone For nodal models the mixing zone results from removing temperature inversions created in the tank by the entering cold water which restricts this region below the inlet position The models can also not represent properly the percentage of heat from the heating element which will be distributed in the mixing zone For these reasons the proportion of heat included in the mixing process will be underestimated as shown schematically in Figure 11 num exp num exp T 0 i 40 30 20 t 2 min t 5 min Figure 11 Schematic evolutions in time of the mixing process comparison between numerical models and experimental measurement 694 Proceedings of Building Simulation 2011 12th Conference of International Building P
8. e in modelling typical residential hot water heater in Qu bec which have two heating elements Therefore TYPE 534 represented the best trade off amongst the 5 models tested in this paper It is also the most flexible in terms of number of inlet ports auxiliary heaters etc REFERENCES Atabaki N Bernier M 2001 Visualisation de la mont e d un panache dans un chauffe eau lectrique Congr s fran ais de Thermique SFT 2001 Nantes 29 31 may 2001 Cruickshank C A Harrison S J 2010 Heat loss characteristics for a typical solar domestic hot water storage Energy and Buildings 42 10 1703 1710 Driick H 2006 Multiport store Model for TRNSYS TYPE 340 Institut f r Thermodynamik und W rmetechnik ITW Universit t Stuttgart Klein S A et al 2010 TRNSYS 17 User s manual Solar Energy Laboratory University of Wisconsin Madison Kleinbach E M Beckman W A Klein S A 1993 Performance study of one dimensional model for Stratified thermal storage tanks Solar Energy 50 2 155 166 SRCC DOCUMENT TM 1 Solar domestic hot water system and component test protocols Solar Rating and Certification Corporation 1679 Clearlake Road Cocoa Florida 32922 5703 www solar rating org July 2008 695
9. e simulation and is controlled by the simulation time step leaving and entering flow rates and node temperatures Both categories of models share most of the same input data geometry material properties etc They offer a different level of flexibility to model specific options and configurations Available options for each studied model are presented in Table 1 688 Proceedings of Building Simulation 2011 12th Conference of International Building Performance Simulation Association Sydney 14 16 November Table 1 TRNSYS models characteristics NODAL PLUG FLOW TYPE 4 TYPE 60 TYPE 534 TYPE 340 TYPE 38 NODES maximum number of nodes 100 100 500 200 45 unequal size nodes y y y INLET MODE load flow enters at the bottom V load flow enters at the specify node y V y y load flow enters at the closest temperature node y y y y y load flow is fractioned and enters at specify nodes HEATING ELEMENTS maximum number of heating elements 2 2 nodes 1 1 internal temperature control y y y y ENERGY BALANCE Flow Streams fully mixed before entering each node y y y y plug flow Thermal losses NI constant overall tank loss coefficient Ll L Aa Hos nodes specified loss coefficient zones top bottom edges specified loss coefficient Heat diffusivity accounts for thermal conductivity fluid tank wall y Temperature inversion instability fully mix appropriate nodes y y Ll 2 aN Lo N Ni use of a mixing flow ra
10. erformance Simulation Association Sydney 14 16 November Mixing zone correction TYPE 534 offers two options to improve the modelling accuracy in the mixing zone e a fractional inlet mode for representing more accurately the entering cold water jet where the water entering the tank is distributed among different nodes according to user set fractions e using a high number of heating elements to represent the uniform heat distribution throughout the mixing zone during water withdrawal Since these fine tuned models have generated better results for a specific water draw profile it wasn t possible to obtain a general rule that will correctly represent all the water draw profile range Result summary Table 4 Assessment of model accuracy according to key criteria in load management TYPE TYPE TYPE TYPE TYPE Cienia 4 60 534 340 38 Supply temperature y x y y o O x N y O V NA vy ov x Energy consumption Operation time Vertical temperature distribution X X X X N A y Correct O Dependent on water draw profile X Significant difference with measurements Table 4 summarizes the results shown above in a qualitative way TYPE 60 results are affected by significant numerical errors that seem to be dependent on the test being performed TYPE 534 and TYPE 340 give the best results overall The vertical temperature profile especially in the mixing zone seems to be a weakness in all the tested models
11. f time for a water draw of 450 and 220 L day Position in Tank cm 20 25 30 35 40 45 50 55 Temperature in Tank C Figure 8 Influence of height and temperature of the mixing zone on the tank supply temperature after a load shifting period for a water draw of 450 L day Power demand Two parameters have been assessed for power demand the time at which the heating element is operating and the total energy use over 24 hours The prediction for the time at which the electrical resistance is activated depends on predicting the thermostat temperature which is approximated by T in the experimental tests This temperature is presented in Figure 9 and will be used to evaluate this aspect Since TYPE 38 does not provide any results on temperatures within the tank element s power demand Qux for this model is used instead Lower thermostat temperature is mainly influenced by the water draw profile due to the water inlet position in the tank For this reason all nodal models are able to predict the time where the lower element will be activated relatively accurately For the plug flow model TYPE 38 some heating events during small water withdraw are missed model heating period Qa does not always correspond to experimental lower thermostat temperature increases 450Liday 80 f EXP TYPE 534 40 70 TYPE 4 TYPE 340 2 TYPE 60 Qaux TYPE38 35 5 ra 5 z 2 z ha oO
12. fected by numerical errors Table 3 24 h overall energy used in kWh by an electrical water tank as a function of tank turnover TYPE TYPE TYPE TYPE TYPE T EXP 4 60 534 340 38 O 198 140 177 177 178 138 08 1283 10 57 1708 10 76 1078 1049 13 1857 15 18 2196 1547 1547 1519 1 7 23 83 19 85 2636 2010 20 10 19 83 Errors occurred at certain time step during simulation Temperature distribution The temperature distribution in the warm zone upper part of the tank is mainly constant and dependent on the thermostat set point Tspr de stratification modelling and tank heat losses The sections above show that it is relatively well modelled by the compared TYPES Figure 10 shows the temperature profile in the so called mixing zone region near the cold water inlet position corresponding to temperature probes Tg To and Tio The results show that under normal operating conditions models that do not account for thermal conductivity de stratification TYPE 4 model a dead zone colder than the experimental results The main advantage of models that do take de stratification into consideration is that over time de stratification effects will correct the temperature distribution in the mixing zone by transferring excess heat from upper nodes which should be located in the mixing zone to the nodes located below In general models overestimate temperature for nodes located above the heating element Tg and un
13. g the inlet position the inlet node was selected to match the position of the end of the internal pipe bringing cold water into the tank Some models allow to maximize stratification by sending the inlet water volume to the node that is the closest in temperature This option has not been used here Thermostat temperature set point The set point temperature for the lower heating element Tser row and the corresponding dead band ATpsg tow have been fixed respectively to 56 5 C and 5 C to match the measured behaviour during the standby test the temperature probe T7 near the lower thermostat was used as a proxy for the thermostat temperature The values for the upper heating element thermostat could not be estimated from measured data so Tsger up was set equal to Tser Low and ATpg up was set at 12 5 C to account for the lowest temperature recorded by the temperature probe T near the upper thermostat These parameters correspond to what can be expected for typical water heaters and default manufacturer settings Heat loss coefficient Some models allow modelling specific U values for each node but this option was not used in order to keep consistent assumptions between the models Two methods were used to calculate the average tank U value using recorded data from the standby test The first relies on the steady state heat loss rate Ug toss and the second relies on the log mean 690 Proceedings of Building Simulation
14. gure 3 Domestic hot water consumption profiles vs average diversified demand Two main tests were carried out e standby test no flow the tank cools down due to thermal losses to the ambient and is heated by the elements in a steady periodic regime e Normal operation test with different hot water draw profiles 220 335 450 L day The discharge profile is regulated by a valve that controls the total volume of hot water leaving the tank at a flow rate of 10 L min The different water draw profiles shown in Figure 3 have been generated in order to approach the average diversified demand of electric water heaters in Qu bec These tests include conditions representative of a simple peak management strategy electric heaters are disabled from 6 AM to 10 AM and 4 PM to 8 PM MODEL PARAMETERS The actual tank geometry was approximated by a cylinder and used in all models Experimental data were then used to determine the other parameters Tank volume segments nodes Based on the relation presented by Kleinbach et al 1993 the recommended number of nodes can be obtained by N nyep 45 8x T 1 where T is the number of tank turnover ratio between the daily water draw and the total tank volume All nodal models were set to use 58 equal size nodes which corresponds to the most stringent conditions in this study average tank turnover 0 8 i e 220 L day 270 L Inlet mode For all models that permitted specifyin
15. have been analysed The next sections report the results of these two tests and also present a comparison of CPU calculation times for the different models Water draw test For this test electric heating elements have been deactivated between 6 AM and 10 AM during the third day of the simulation At the beginning of each hour 11 L of water are drawn from the tank with a flow rate of 10 L min Figure 4 shows the temperature in the tank horizontal axis versus the height in the tank vertical axis at different times of the day different labelled curves Results show that the models taking de stratification into account give similar temperature profiles TYPE 4 is different since it does not model de stratification due to thermal conduction which results in a dead zone in the bottom part of the tank where the entire lower portion stabilizes at 22 C which is the ambient temperature It is also worth noting that the different results for TYPE 60 at the top of the tank are due to numerical errors that only show up in some circumstances In this particular instance our tests revealed that the errors would go away if a lower water draw flow rate e g 2 L h was used in the simulation Nevertheless these results show that TYPE 60 is not as robust as other models tested in this particular case TYPE4 TYPE 60 Position in Tank cm g PI ai _ TYPE 534 TYPE 340 Position in Tank cm eo oO 60 25 30 35
16. i et al 2001 TYPE4 TYPE 60 10 00 am 10 00 Am 420 10 15 AM 10 15 AM 10 30 AM i 10 30 AM 100 10 45 AM ea eo loses 10 45 am Position in Tank cm o TYPE 534 TYPE 340 10 00 am 10 15 Am 10 30 AM 10 45 Am 10 00 am 120 10 15 AM 10 30 Am 100 10 45 am Position in Tank cm o 20 25 30 35 40 45 50 55 60 0 25 Temperature in Tank C Temperature in Tank C Figure 6 Tank vertical temperature distributions during an electric heater charging phase Computational efficiency Table 2 presents the total calculation times for each model for a 72 h simulation with a 15 sec time step The results are based on values optionally reported by TRNSYS for each TYPE In the case of TYPE 534 calculation time for TYPE 1502 temperature controller is also accounted for since all other components perform the temperature control inside the TYPE This contribution is very small lt 2 TYPE 38 Plug flow is by far the less computationally intensive mainly because of the fewer nodes i e plugs of fluid used to model the tank due to operating conditions the component typically uses less plugs than the maximum permitted by the model 45 at any given time All other models use the same number of nodes but they do show significant differences in computational efficiency TYPE 340 and TYPE 60 are significantly faste
17. ile experimental results show a significant temperature drop at the end of the off period which is not reproduced by the models The behaviour of the models represented by the TYPE 534 and the differences with the experimental data are shown in Figure 8 the green line corresponds to the time of the second dip in Figure 7 The model predicts a step like transition between a cold zone and a warm zone while in reality the bottom of the tank is occupied by a slightly warm mixing zone and the top of the tank is not as warm as modelled The inability of the models to represent the mixing zone accurately will be discussed further below The plug flow model shows a different behaviour and its performance is actually worse for smaller tank turnover ratios lt 0 8 The turnover ratio has an impact on the number of nodes plugs used in the model and it seems that low water draw flow rates result in using a number of plugs that is insufficient to obtain a good accuracy This is shown in the lower part of Figure 7 for TYPE 38 450L day 60 D z a W H 2 5 52 EXP TYPE 534 TYPE 4 TYPE 340 TYPE 60 TYPE 38 50 T 220 L day 60 o D gZ i i 54 i ke i i i H 5 5 52 ie po oOo a EXP TYPE 534 J TYPE 4 TYPE 340 TYPE 60 TYPE 38 50 54 60 66 72 Time hour Figure 7 Hot water supply temperature as a function o
18. le 270 L standard electric water heater storage tank This glass lined steel tank is commonly used in 689 Proceedings of Building Simulation 2011 12th Conference of International Building Performance Simulation Association Sydney 14 16 November presented in Figure 2 In regions away from the thermocline the wall temperature is slightly lower 0 1 C than the actual water temperature indicating that there exist a temperature gradient in the tank wall resulting from heat losses to the ambient Thus temperature probes underestimate the real water temperature Furthermore this effect is amplified at both tank extremities where the contact area with the environment is larger Near the thermocline the error introduced by measuring the external wall temperature instead of the water temperature is significant The affected zone is approximately 3 5 cm in height Based on this analysis a combined measurement uncertainty of 1 C was assumed for temperature measurements to account for this effect and for thermocouple measurement uncertainty 0 5 C Water Tank exterior wall Position in Tank cm 40 50 60 Temperature C Figure 2 Simulated water and tank temperatures for a 1 D no flow pure conduction case Experimental procedure 35 T T T T T T T T T T T MM 225 L day ME 330 L day ME 450 L day Avg power profile Water consumption L a Power consumption kW 6 12 18 24 Time hr Fi
19. lect the most appropriate model among the ones available in TRNSYS in the context of assessing demand side management scenarios for residential domestic water heaters COMPARED MODELS The storage tank models considered for this study were selected for their ability to represent a typical electrical water heater i e a vertical cylindrical tank with two heating elements Five TRNSYS models referred to as Types were identified e three Types from the standard TRNSYS library TYPE 38 plug flow tank model TYPE 4 and TYPE 60 two different stratified tank models based on a nodal approach e one TYPE from the TESS component libraries TYPE 534 detailed stratified tank using a nodal approach e one non standard component distributed by Transsolar TYPE 340 multi port water storage using a nodal approach It should be noted that TYPE 38 and TYPE 340 can only model one heating elements but they were considered in this study the results below will show that the upper element is never used in the experimental and intermodel comparison tests The selected models can be organized in two main categories They will be referred to the nodal and plug flow approach In the nodal approach the tank is modelled by a fixed number of nodes control volumes that hold a fixed amount of water at a fixed height in the tank In the plug flow approach the tank nodal segmentation varies during th
20. ll 50 W mK corresponding to Kgee Ac tank wal 0 006 m is the cross sectional area of the storage wall and Agwater 0 185 m is the cross sectional area of the water in the tank This gives Kerr 2 25 W mK Simulation time step Some models imposed restrictions on the simulation time step Although this problem is not clearly documented TYPE 4 will suffer from significant errors in the tank energy balance if the fluid volume of any of the nodes is completely replaced within a given time step in other words if the flow rate through a node is equal to or higher than the volume of the node divided by the time step Given the maximum flow rate in our case is 10 L min and the node volume of 4 66 L node has a capacity of 270 L 58 nodes the maximum time step is 28 sec A value of 15 sec was selected for all intermodel tests and for comparison with experimental data INTERMODEL COMPARISON The series of intermodel comparison tests aimed at comparing the ability of different models to calculate the vertical temperature profile in the tank Two different tests are performed first profiles are compared during and immediately after a water draw event with all electric heating elements deactivated then during a charging phase with the bottom electric heating element activated without any water draw in the tank To minimize the impact of different initial tank conditions on the results the last 24 hours of a 72 hour simulation
21. ng makes a significant contribution to the electrical grid peak load and it represents an opportunity for peak shaving However load shifting may have some detrimental consequences on the domestic hot water supply temperature if the heating element is deactivated for a long period Furthermore a new peak may be caused if a significant number of heaters are reactivated at the same time Accurate models are required to assess the impact of load management on residential domestic hot water heaters The local standard residential electric water heaters consist of a cylindrical tank with typical nominal capacities of 180 L or 270 L Hot water leaves by the top and cold water enters directly in the bottom or is brought from the top by an internal pipe Water is heated by two horizontal elements with a rated power ranging from 3 to 4 5 kW These elements are regulated by two thermostats in a master and slave mode where the upper element as the priority and the lower element can only be activated if the upper one is off In practice the upper and lower heating elements often have the same set point e g 60 C but the upper element has a larger deadband 10 C than the lower element 5 C This results in the lower element being activated most of the time even though it has a lower priority in theory OBJECTIVE The purpose of this study is to assess the level of modelling detail that is required to obtain acceptable predictions and to se
22. r The results also show that the tank turnover T has a significant impact for TYPE 534 on calculation time for a given time step Table 2 Calculation time inside TYPES for different number of tank turnover for a 72 h simulation sec TYPE TYPE TYPE 534 TYPE TYPE q 4 60 TYPE1502 340 38 0 21 72 5 79 18 80 4 92 0 35 0 8 20 94 6 40 20 37 4 58 0 31 1 3 22 33 6 42 25 39 4 44 0 35 1 7 22 80 6 53 30 22 3 90 0 36 EXPERIMENTAL VALIDATION The experimental validation focuses on 3 main performance aspects that could be affected by demand side management e Domestic hot water DHW supply temperature related to user comfort e Power demand time dependent profile and overall energy use 692 Proceedings of Building Simulation 2011 12th Conference of International Building Performance Simulation Association Sydney 14 16 November e Temperature profile in the tank e g control of bacterial contamination Hot water supply temperature Model predictions for supplied hot water temperature have been compared with the measured data during the last simulation day As shown in Figure 7 for all models with a nodal approach except for TYPE 60 for which numerical errors were encountered the outlet temperature predicted is within the experimental uncertainty margins 1 C The differences can become larger if the heating element is disabled for long periods with a high tank turnover ratio For the 450 L day water draw prof
23. te y EXPERIMENTAL PROCEDURE residential domestic hot water systems The electric EXPERIMENTAL PROCEDURE heater elements are both rated at 4 2 kW The cold Description of the experimental set up water inlet is at the top of the tank but an internal pipe brings the inlet water to the bottom approximately at the height of the lower heating element thermostat Flow meter For practical reasons the vertical temperature distribution has been measured by temperature probes installed on the storage wall between the steel tank and the insulation layer Figure 1 shows the location of the temperature measurements green dots thermostats and heating elements the separating distances on the left and right are given in cm The inlet flow rate is measured by a turbine flow meter and the power used by each heating element is measured independently A data acquisition system collects data at a 5 min time interval ka Q AUX Upper fi Q aux lower 20 ak The influence of using the tank wall temperature measurement instead of a water temperature measurement inside the tank was investigated numerically in COMSOL COMSOL Multiphysics 3 5a The numerical calculations assume a no flow pure 1 D conduction problem with a sharp temperature gradient in the tank thermocline and typical tank insulation Results of this analysis are Tho lobbbelal a Ke Figure I Experimental set up Experimentation was made on a commercially availab
Download Pdf Manuals
Related Search