General comparison of models - Comparison of level of detail in simulation models

5.3 Comparison of level of detail in simulation models

5.3.5 General comparison of models

The comparison starts by a general overview of the results from the simulations with focus on the energy consumption. In the following sections, the differences are analysed further to find both the differences in the models but also to generally analyse floor heating systems and numerical modelling of floor heating systems.

Energy flows

Figure 5.31 shows the energy flows in the simulation model for one of the simulation models with heat recovery unit. The passive gains are solar gain and the internal heat gain while the

“heat to pipes” represents the heat used to heat the building through the floor heating system.

The losses are split into transmission losses through walls and ceiling, windows and ground;

while ventilation, infiltration and venting represent the airborne heat losses. The values are shown for the heating season, which in the Danish Design Reference Year is defined to be from September 24 to May 13.

−1000 −500 0 500 1000 1500

solar income internal heat gain heat to pipes transmission: walls transmission: windows ground heat loss infiltration ventilation venting

Energy [kWh]

Energy flows during heating season

Figure 5.31 Energy flows in model shown for the 2D model with heat recovery unit on the ventilation.

Energy consumption and heat loss to the ground

Figure 5.32 and Figure 5.33 show the energy consumption for the seventeen different variations of the floor model with the three different insulation thicknesses. A supply temperature of 35°C is used in the floor heating pipes. The first figure shows the results for the model without heat recovery unit which has the largest energy consumption while the second figure shows the results with heat recovery unit on the ventilation air.

0 10 20 30 40 50 60 70 80 90 100 Electrical 1D

Electrical 1D with ground Electrical 1D (ISO13370) 1D with pipe 1D with pipe and ground 1D with pipe (ISO13370) 1.5D with pipe 1.5D with pipe and ground 1.5D with pipe (ISO13370) 2D large ground volume 2D no floor heating RC−model RC−model with psi−value RC−model (ISO13370) Serial 1.5D Serial 1.5D with ground Serial 1.5D (ISO13370)

Energy consumption [kWh/m²] − T_supply = 35 °C − No HRU − Heating season 150 mm insulation

250 mm insulation 350 mm insulation

Figure 5.32 Simulated energy consumption using the different simulation models without heat recovery unit on the ventilation system.

0 5 10 15 20 25 30 35 40 45

Electrical 1D Electrical 1D with ground Electrical 1D (ISO13370) 1D with pipe 1D with pipe and ground 1D with pipe (ISO13370) 1.5D with pipe 1.5D with pipe and ground 1.5D with pipe (ISO13370) 2D large ground volume 2D no floor heating RC−model RC−model with psi−value RC−model (ISO13370) Serial 1.5D Serial 1.5D with ground Serial 1.5D (ISO13370)

Energy consumption [kWh/m²] − T_supply = 35 °C − Heating season 150 mm insulation

250 mm insulation 350 mm insulation

Figure 5.33 Simulated energy consumption using the different simulation models with heat recovery unit on the ventilation system.

The energy consumption is around 80kWh/m² for the model without heat recovery unit, while it is around 30kWh/m²-35kWh/m² for the model with heat recovery unit. The results are most uniform for the high energy consumption regardless of model type and insulation thickness.

For the model with low energy consumption the differences are larger relative to the total energy consumption.

The smallest energy consumption in both figures are for the ground coupled 2D model

without floor heating, where the heat is supplied directly in the room air zone, which will give a lower energy consumption as the temperature in the room is controlled perfectly to the set point temperature. The lowest values with floor heating are found in the models with the electrical inclusion of the floor heating system. This is due to the fact that electrical floor heating has faster reaction time than using hydronic systems. Also notice, that the basic RC-model has lower energy consumption than the other RC-models. The highest energy consumption is found in the 2D model with large ground volume and in the RC-model with the linear thermal transmittance included. This is a natural consequence of the inclusion of the foundation in these two models. The models based on ISO13370 have smaller energy consumption than these two models, even though they also have the foundation and ground volume included.

The energy consumption found in the 1D and 1.5D models is almost the same when comparing the three different models from each group. The serial 1.5D models can also be seen to give almost the same results as the 1.5D model which only uses a single model.

The difference in the energy consumption when the insulation thickness is changed is also shown in the figures. As expected, all models have smaller energy consumption when the insulation thickness is increased, but the rate of decrease is very different in the models. All models based on EN ISO13370 have small differences in the energy consumption with the different insulation thickness compared to the other models in the same group. This is also the case with the 2D model without floor heating. As opposed to this, the difference for the 2D model with floor heating and the RC-model is much larger when the insulation thickness is changed. This is especially the case when going from 150mm insulation to 250mm insulation with low energy consumption.

It can also be noticed that comparing the results in the models for the different insulation thicknesses gives very different results. As an example the 2D model with floor heating and the RC-model with linear thermal transmittance can be compared in Figure 5.33. For an insulation thickness of 350mm the energy consumption is almost the same for both, while the results are very much different for 150mm insulation thickness.

Looking at the different insulation thicknesses in Table 5.12 individually – the average, maximum and minimum values of the simulated energy consumption for the models with floor heating – very large differences can be found. For the model without heat recovery, the difference is almost 11-15kWh/m². For the model with heat recovery it is more stable around 6-8kWh/m².

Table 5.12 Average, maximum and minimum value of the energy consumption in the 16 different floor construction models with floor heating.

Insulation thickness

[mm]

Average [kWh/m²]

Maximum [kWh/m²]

Minimum [kWh/m²]

150 86.8 95.0 80.0 250 82.8 88.4 78.5 Without heat

recovery

350 80.8 85.5 76.6 150 35.2 40.2 32.2 250 32.3 36.2 29.3 With heat

recovery

350 30.8 34.4 28.0

In Figure 5.34 and Figure 5.35, only the heat loss to the ground is shown. Here the differences are more obvious than for the total energy consumption. It especially becomes obvious how the insulation thickness influences the heat loss to the ground.

0 5 10 15 20 25 30 Electrical 1D

Heat loss to the ground [kWh/m²] − T_supply = 35 °C − No HRU − Heating season 150 mm insulation 250 mm insulation 350 mm insulation

Figure 5.34 Simulated heat loss to the ground using the different models for models without heat recovery unit on the ventilation system.

0 5 10 15 20 25 30

Heat loss to the ground [kWh/m²] − T_supply = 35 °C − Heating season 150 mm insulation 250 mm insulation 350 mm insulation

Figure 5.35 Simulated heat loss to the ground using the different models for models with heat recovery unit on the ventilation system.

The heat loss to the ground for the models without foundation and ground volume is lower than the models where this is included. Generally the differences between the models are smaller for larger insulation thicknesses. Models based on EN ISO 13370 again have smaller difference between the heat losses for different insulation thicknesses.

Comparing the 2D models with and without floor heating, it can be seen that floor heating have larger heat loss to the ground, which is as expected since the floor temperature is higher.

Generally the heat loss to the ground in the model without floor heating is around 85% of the heat loss with floor heating in the model with low energy consumption, while it is around 75% for the model with high energy consumption. This means that that since the floor is not heated as much for the low energy consumption as it is for the high energy consumption, the relative difference in the ground heat loss is also smaller for the house with low energy. At the same time it has been found that the relative difference is independent of the supply

temperature to the floor heating system. This is not shown in the figures.

It is noticed, that even though the total heat loss is more than twice as large for the model without heat recovery unit on the ventilation system, the heat loss to the ground is almost the same, differing by less than 10% in most cases. For the models based on ISO13370, the heat loss to the ground is the same both for the model with high and low energy consumption, since the average concrete temperature is not calculated separately in the two models.

Similar to Table 5.12, Table 5.13 shows the average, minimum and maximum values of the ground heat loss for the models with floor heating. Especially for the small insulation thickness it is obvious that the models have large differences, where the minimum value is only about half of the maximum value.

Table 5.13 Average, maximum and minimum value of the heat loss to the ground in the 16 different floor construction models with floor heating.

Insulation thickness

[mm] Average

[kWh/m²] Maximum

[kWh/m²] Minimum

[kWh/m²]

150 24.2 35.8 18.3 250 18.2 25.7 16.7 Without heat

recovery

350 15.0 20.9 13.2 150 22.1 31.6 18.3 250 16.8 22.9 15.4 With heat

recovery

350 14.0 18.6 11.8

Finally, the ratio of the heat loss to the ground to the total energy consumption has been calculated for the models. This value is 10% to 20% for the model with large energy consumption, while the corresponding values are 25% and 50% for the model with low energy consumption; largest for the model with the low energy consumption. A heat loss of up to 50% through the ground and foundation of the total energy consumption in new buildings with low energy demand is comparable to findings in other works as described in Chapter 3.

Dynamic response for the different models

The response is compared through the time constant. In this case, the flow is turned on from a steady state situation where the room is at 21°C and the ground is at 10°C. The pipe

temperature is set to 35°C. The time constant – which has been defined in Chapter 3 as the time it takes for a step change to reach 1−e⁻¹63.2% of the final value when the unit step change is applied as input – is almost the same in all cases ranging from 4 to 5 hours, with the RC-model being the fastest and the serial models being slowest of the models with pipe implementation. The electrical implementation has a slower reaction time because the heat output from the “pipe” is constant once it has been turned on. This is not the case in the hydronic models, where the maximum heat output peaks right after the flow has been turned on and drops towards a steady-state value hereafter. It should be noticed that the time constant very much depends on the value of the thermal resistance between fluid and concrete.

However, this value is very difficult to estimate. In fact, by changing the value from a “low”

to a “high” value, the time constant for the 1D model changes from 3.5 to 5 hours. However, the energy consumption only changes by about 4% in a yearly calculation. Because of the lack of a better value, the value in the 1D model is set to 0.1m²K/W between the pipe and concrete. The thermal resistance between pipe and concrete deck is therefore important for finding the time constant, while it is not important with respect to the energy consumption.

Figure 5.36 shows the operative temperature for all models. One curve stands out; namely the one representing the model without floor heating. Because of the lower radiant temperature, the operative temperature is lower than for the models with floor heating. Generally the temperature can be controlled to above 21°C, which is the set point temperature for the room air temperature. The models have nearly the same peaks during periods with solar gains, differing by less than 1 K. Without solar gains during an entire 24 hour period, as is the case between approximately hour 1475 and 1500, the models have different temperature courses.

1350 1400 1450 1500 1550

20 21 22 23 24 25 26

Time [hours]

Temperature [°C]

Operative temperature in the 17 models

Figure 5.36 Operative temperature in the 17 simulation models. It is not indicated which temperature profile belongs to which model, as it would not be possible to make distinctions between them.

During the summer period, the temperature profiles are almost identical, regardless of type, indicating that when the heating system is turned off the models have almost the same behaviour, which is shown in Figure 5.37.

4700 4725 4750 4775 4800 4825 4850 4875 4900

20 21 22 23 24 25 26 27 28 29 30

Time [hours]

Temperature [°C]

Operative temperature in room

Figure 5.37 Operative temperature during summer period.

Supply temperature

Figure 5.38 and Figure 5.39 show the energy consumption with 250mm insulation for three different supply temperatures, 35°C, 45°C and variable, as shown in Figure 5.30. The models with electrical floor heating and the 2D model without floor heating have the same energy consumption regardless of the supply temperature, as this parameter is not used as input. In almost all cases, the energy consumption is lowest for the variable supply temperature and highest with 45°C. This is as expected and in accordance with the description in Chapter 3.

0 10 20 30 40 50 60 70 80 90 100

Energy consumption [kWh/m²] − 250 mm ground insulation − No HRU − All year Tsupply=35 °C

Tsupply=45 °C Tsupply variable

Figure 5.38 Simulated energy consumption for the different floor models with different supply temperatures – here shown for 250mm insulation without heat recovery on the ventilation air.

0 5 10 15 20 25 30 35 40 45 50

Energy consumption [kWh/m²] − 250 mm ground insulation − All year T_supply=35 °C

Tsupply=45 °C Tsupply variable

Figure 5.39 Simulated energy consumption for the different floor models with different supply temperatures – here shown for 250mm insulation with heat recovery on the ventilation air.

A very relevant discussion when comparing the calculated energy consumption with different supply temperatures is that the thermal comfort has not been the same. Therefore, a model with larger energy consumption will have higher average temperatures. However, this will not necessarily lead to a higher level of thermal comfort, as the higher temperature is a

consequence of more inaccurate control of the indoor temperature. To assess the thermal comfort, a duration curve or temperature distribution curve during the heating season can be used. Figure 5.40 shows this curve for model type 7 using three different supply temperatures;

35°C, 45°C and variable. The curve shows the number of hours during the heating season below a given operative temperature. For instance, there are approximately 2000 hours during the heating season when the operative temperature has been 22°C or less for the case with 45°C supply temperature in the right figure. The figures show the temperature distribution during the heating season from Sep. 24 to May 13, approximately 5500 hours.

The left figure shows the model with high energy consumption, while the right figure shows the model with low energy consumption. In all cases, there are very few hours with

temperatures below 20°C or even 21°C, which is satisfactory seen from a thermal comfort point of view. However, especially for the right figure, it is obvious that the supply

temperature has a large impact on the distribution of the operative temperature during the heating season, as there is a large difference between the number of hours below 22°C with 2000 hours with 45°C supply temperature and just over 3000 hours with the variable supply temperature. This means that there are many hours with too high temperatures for 45°C supply temperature. As it can be seen in Figure 5.39, the difference in energy consumption from the variable supply temperature to 45°C is approximately 10%.

The right figure shows a more even distribution of the operative temperature in the room, which can also be seen in Figure 5.38, as the models differ less in the energy consumption, though with a larger value for 45°C than for the other two. Therefore the results for a building with low energy consumption are more influenced by a high supply temperature.

18 20 22 24 26 28 30

0 1000 2000 3000 4000 5000 6000

Operative temperature [°C]

Hours below [hours]

Insulation: 350 mm. Modeltype: 7 − No HRU

T_supply=35°C T_supply=45°C T_supply variable

18 20 22 24 26 28 30

0 1000 2000 3000 4000 5000 6000

Operative temperature [°C]

Hours below [hours]

Insulation: 250 mm. Modeltype: 7 − With HRU

T_supply=35°C T_supply=45°C T_supply variable

Figure 5.40 Operative temperature distribution curves for three different supply temperatures, based on the 1.5D model without ground volume. The data with high energy consumption are shown to the left.

If the curves are compared for the two models, notice that the model with low energy

consumption (the right figure) has a sharper curve around 21°C than for the model with high energy consumption. This indicates that the control is more efficient, as there are basically no hours with low temperature. Further, the heating system can easily supply sufficient heat to ensure that the set point temperature can be upheld. For the model without heat recovery unit there are around 100 hours with temperatures below 21°C for the variable and 35°C supply

temperature. The upper part of the distribution curves (above 24°C) shows almost the same behaviour for both high and low energy consumption, a consequence of the fact that for high temperatures the solar gains almost exclusively decide the temperature in the room.

Room temperatures with and without floor heating

The temperatures in the room are very different comparing the model with and without floor heating. Figure 5.41 shows three different supply temperatures (35°C, 45°C and variable) along with the model without floor heating. The set point temperature is an air temperature of 21°C. The most obvious difference is the floor surface temperature, which is much lower for the model without floor heating. Here the lowest temperature is 19.5°C compared to 21°C for the models with floor heating. Notice that 19°C is still above the lower comfort limit for floors as described in Chapter 3. It can also be seen how large the influence from the lower floor surface temperature is on the mean radiant temperature, as the number of hours below any temperature is considerably larger than when floor heating is present.

18 20 22 24 26 28 30

0 1000 2000 3000 4000 5000 6000

Air temperature [°C]

Hours below [hours]

Insulation: 250 mm − With HRU

Tsupply = 35°C Tsupply = 45°C Variable supply temp without floor heating

18 20 22 24 26 28 30

0 1000 2000 3000 4000 5000 6000

Floor surface temperature [°C]

Hours below [hours]

Insulation: 250 mm − With HRU

Tsupply = 35°C Tsupply = 45°C Variable supply temp without floor heating

18 20 22 24 26 28 30

0 1000 2000 3000 4000 5000 6000

Operative temperature [°C]

Hours below [hours]

Insulation: 250 mm − With HRU

Tsupply = 35°C Tsupply = 45°C Variable supply temp without floor heating

18 20 22 24 26 28 30

0 1000 2000 3000 4000 5000 6000

Mean radiant temperature [°C]

Hours below [hours]

Insulation: 250 mm − With HRU

Tsupply = 35°C Tsupply = 45°C Variable supply temp without floor heating

Figure 5.41 Comparison of temperature distribution of air, floor surface, operative and mean radiant temperature for model type 10 and 11 with and without floor heating respectively

The function of the heat supply to the model without floor heating can be seen in the upper left part of Figure 5.41 for the air temperature where there is a straight line at 21°C with 0 hours below this limit.

The maximum room air temperature is 26°C is achieved by increasing the ventilation rate to ensure that the temperature reaches 26°C.

It is very important to notice that this comparison to an “ideal” heating system is not realistic, because of the way the heating is supplied to the room – namely by a perfect control which ensures that the room air temperature is always precisely 21°C when heating is required. Such a heating system would need to have an infinitely fast reaction time, which is of course not possible. A heating system with radiations placed under the windows on the outer walls will also not be able to control the temperature accurately at the desired set point and will due to its higher temperature also give a higher heat loss through the outer wall and window.

Comparisons of different heating systems have been carried out in papers and reports, mentioned in Chapter 3. Here it is only important to notice that model type 10 with “ideal”

heating is not possible.

Simulation time

The simulation time for the program execution is shown in Figure 5.42. The 2D models are by far the most time consuming. The 1D and RC models are approximately 10 times faster, while the 1.5D models are about 7 times faster and the serial 1.5D model is around 5 times faster.

The simulation time in FHSim for the 2D model is approximately 45 minutes on a 3.4 GHz Pentium 4 processor for one year of simulation. The fastest models have a simulation time of approximately 5 minutes. Notice also that the 2D models typically require five years or more of simulation before the conditions in the ground volume are periodic stationary, which further means that a long extra simulation time is needed for these models.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Simulation time relative to longest time

Figure 5.42 Relative simulation time for entire simulation program. The longest simulation time is approximately 45 minutes.

Figure 5.43 shows the simulation time for the models relative to the 2D model (type 10), for the floor model only. The rest of the simulation program has not been included. Notice, that data is shown in a logarithmic scale because of the large differences in the simulation time.

The simulation time varies by a factor of over 100 between the fastest RC thermal network models (type 12-14) to the 2D models (10-11). The 1D models (type 1-6) have almost the

In document Modelling building integrated heating and cooling systems (Sider 114-124)