3.5 Task 1: Model development and validation
3.5.10 Model validation
This section describes the processes employed to validate the models used in this work. Two methods have been used to verify that the model development activities and the results generated are plausible and represent reality: 1) inter model performance comparison, which involved simulating and comparing the thermal and energy performance of a 1990s dwelling in IES Virtual Environment (IES-VE) and TRNSYS, and 2) comparing the simulated thermal and energy performances of the dwellings considered in this work with published results of thermal and energy performance measurement of similar dwelling with similar occupancy and operational conditions.
The purposes of these exercises were to ascertain whether the models, when reproduced in a different dynamic energy and building modelling software and simulated with similar sets of input data, produces the same results. A further purpose was to ensure that the output generated by the models are realistic and plausible given the various assumptions which are necessary in this type of work, and the estimated nature of certain simulation parameters such as the weather data. Ultimately the cross checking of the results generated in two different modelling software and the comparison of the predicted energy consumption data with published monitored data (for buildings of similar sizes and operational conditions) showing similar results provided confidence in the modelling and the simulation processes followed during the research, and the results obtained as described in the results chapters.
Inter model comparison
Inter model comparison was carried out by means of creating and simulating models in TRNSYS and IES-VE of a building with identical construction, occupancy and operational conditions. The assumptions and methods used in creating the building models and the heating system components are identical to those described in
Thermal energy storage in residential buildings: A study of the benefits and impacts
124
sections 3.3.4 to 3.3.7. The thermal and energy performance were simulated in both modelling tools and the results compared (see Figure 3-14).
TwbSYS Model LES-VE Model TwbSYS Output LES-VE Output Comparison of output Model parameters
Heating type & size Occupancy Boundary Conditions Geometry data Weather data Construction data Lnternal gains Simulation control wesolution Simulation period
Model & simulation control adjustment
Figure 3-14. Inter model validation block diagram comparing the simulated energy and thermal performance of a dwelling with identical construction, occupancy and operational conditions in TRNSYS and IES-VE
The physical properties of the building modelled here was different to the building models described in Section 3.5.4, and is based on a real detached test house building located on the Loughborough University campus. The building is of standard brick construction and built to the 1990’s building regulation standards. The construction parameters of the main building components are as described in Table 3-8. One occupancy schedule (occupancy Type B as previously described in Figure 3-8) and heating system operational schedule corresponding to a 9 hour heating duration (7:00 to 09:00 and 16:00 to 23:00) with a thermostat setting of 21℃ was used. The internal gain parameters utilised was as described in Table 3-5. The air infiltration level was set to 1 ACH and only applied during the daytime occupied hours.
a) b)
Thermal energy storage in residential buildings: A study of the benefits and impacts
125
The building model was initially created in IES-VE (see Figure 3-15a) with the build and construction details extracted from the original drawings. Parameters, which were not present on the drawings, were based on the standard specifications of the 1990’s building regulation requirements. A TRNSYS model of the same building was created using a Type-56 multi-zone building model (see Figure 3-15b), and was adapted to include construction materials with identical or closely matching thermal and physical properties to those used in IES-VE, thus possessing similar overall thermal characteristics. For example, the external wall representation in IES-VE and TRNSYS had overall U-values of 0.715 W/m2K and 0.702 W/m2K respectively.
Table 3-8. Building fabric construction details used in the IES-VE and TRNSYS models.
U-value (W/m2K) Building material & thickness (m) Ground Floor 1.39 0.01 Synthetic carpet 0.05 Rubber underlay 0.02 Timber flooring 0.30 Air gap 0.15 Cast concrete 0.15 Stone chipping 0.75 London clay
External Wall 0.72 0.100 Facing brick outer
0.058 Cavity insulation 0.150 Block inner skin 0.015 Plaster finish Internal Partition
Walls 1.59 0.150 Gypsum plasterboard 0.100 Cavity
0.150 Gypsum plasterboard
Roof 4.87 0.025 Slate tiles
0.01 Ashfelt External Windows
PVC 1.98 Double glazed 0.006 Pilkington glass
0.012 Cavity 0.006 Pilkington glass
Internal Doors 2.29 0.04 plywood
External Doors 2.55 0.047 Oak door
FF floor/GF ceiling 1.28 0.01 Synthetic carpet
0.05 Rubber underlay 0.02 Timber flooring 0.300 Cavity
0.150 Gypsum plasterboard
Loft floor/FF ceiling 0.22 0.170 Insulation
0.150 Gypsum plasterboard
Model configuration and simulation control
The heating system in IES-VE was modelled within the ApacheHVAC modeller, as water based central heating system with convector radiators adding up to 10kW, connected to a 20kW generic electrical heat source. On/Off controllers were used to control the heating. The heating system also provided the domestic hot water. The TRNSYS heating system configuration was as described in Section 3.5.7.
Thermal energy storage in residential buildings: A study of the benefits and impacts
126
The weather data used in TRNSYS and IES-VE are based on two locations; 1) approximately 5 miles north, and 2) approximately 15 miles north of the building location respectively. Simulations were performed for 60 days from 1st of January. It can be appreciated that the use of two separate weather data files, although they are based on virtually the same geographical location, have some inherent differences, and therefore the two simulations cannot be considered identical. In any case, it would be wrong to suggest that the IES-VE model, even with identical parameter values but without it being validated first with some scientifically acceptable means, could be used to validate the TRNSYS model. However, as stated earlier, IES-VE is a commercial package with tightly controlled modelling environment, that has been tested using ASHRAE Standard 140 and qualifies as a Dynamic Model in the CIBSE system of model classification (Crawley et al. 2005), and widely used both in industry and academia. With the due care and attention, it can be used to generate acceptable thermal and energy performance results of simple buildings, such as the one used in this exercise, relatively easily and with minimal uncertainty. With this in mind, this exercise was carried out with the aim to: a) see whether there is a trend wise agreement between the results produced by the two models in the short time scale, and b) see whether there is agreement between the results over the long term, where the short term differences, for example the day to day differences in the external ambient temperature, average out to a certain degree.
Comparison of the simulation results
Figure 3-16 and Figure 3-17 illustrate the occupied space temperature, the external ambient temperature and the heating load curves for an example day of 7th January for both the TRNSYS and IES-VE simulations. The main parameters simulated by the two applications are summarised in Table 3-9 for the example day and for the 60
day simulation period from the 1st of January. It can be seen that there is
considerable trend wise agreement between the responses generated by the two applications. The space temperature transient response are similar, both having a rise time to the thermostat setting of around 35 to 40 minutes from the start of the heating system. The space temperature fluctuates around 21℃ as the heating system cycles on and off with the space temperature rising and falling outside a 2℃ dead-band, centred at the thermostat setting of 21℃. The heating system energy
Thermal energy storage in residential buildings: A study of the benefits and impacts
127
demands during the example day were 67.1kWh and 65.7kWh for TRNSYS and IES- VE respectively. The difference between these values could be due to the slightly lower external ambient temperature in the TRNSYS weather data (see Figure 3-17) requiring more heat to maintain the space temperature.
Figure 3-16. Example of the thermal response on 7th January from the IES-VE.
Figure 3-17. Example of the thermal response on 7th January from the TRNSYS model.
The 60 day average space temperature and the heat energy demand predictions (for space heating only, DHW only and combined total) are similar for both IES-VE and TRNSYS. The noteworthy differences are that the TRNSYS predictions for the space heating energy demand is marginally higher; the difference between the maximum and the minimum daily space heating energy demand prediction over 60 days is slightly bigger, and the mean and the minimum occupied space temperatures are
0 5 10 15 20 25 0 5 10 15 20 25 0 2 4 6 8 10 12 14 16 18 20 22 H ea ti n g L o ad ( k W ) R o o m T em p er at u re ( °C) Time GF_Temp Amb_Temp Heat&DHW_Load_W
Thermal energy storage in residential buildings: A study of the benefits and impacts
128
marginally lower (see Table 3-9). However these could be explained by the slightly poorer external ambient conditions inherent in the TRNSYS weather data, for example over the 60 day simulation period the maximum, minimum and the mean external ambient temperature in the TRNSYS weather data being 12.2℃, -5.0℃ and 3.4℃ respectively. The same figures in the IES-VE weather data are 12.4℃, -2.9℃ and 4.29℃ respectively. Therefore it is reasonable to expect a marginally higher average heating energy demand and a marginally lower space temperature predictions (outside of the heating periods) in TRNSYS compared to those of IES- VE. Some differences were also expected due to the minor variations in the building fabric models used in the two simulation packages as previously explained, for example the external wall U-value in the in IES-VE and TRNSYS models being 0.715 W/m2K and 0.702 W/m2K respectively.
In summary, the differences in the predictions generated by the two simulation tools mentioned are relatively small, for example the difference between the space heating energy demand prediction by TRNSYS and IES-VE over the 60 day simulation period is 2.6%. Further analysis of the results indicated good correlation between the weather data parameters, such as solar radiation and external ambient air temperature, and the space temperature responses and the heat energy demands predictions generated by the two applications. For example the heat demand and the minimum room temperatures varied correspondingly with the external ambient temperature and the solar radiation levels.
Table 3-9. Comparison of the key heat consumption and space condition predictions by IES-VE and TRNSYS for a sample day of 7th January and for a 60 day simulation period from 1st January.
TRNSYS IES-VE
7th January 60 days from
1st January 7th January 60 days from 1st January
Mean space temperature (℃) 15.6 15.7 17.1 17.5
Minimum space temperature (℃) 9.7 9.7 12.4 10.79
Maximum space temperature (℃) 22.0 22.0 22.1 22.3
Space heating energy (kWh) 57.8 55.3 56.2 54.3
DHW energy (kWh) 9.2 9.2 9.4 9.4
Total energy (kWh) 67.1 65.4 65.7 63.7
Maximum total heat energy (kWh) - 81.5 - 76.5
Minimum total heat energy (kWh) - 48.0 - 49.4
Validation through comparison of published measured data
Further validation of the model performance and the results generated was carried out by comparing the simulated heating energy consumption prediction with published measured thermal and energy performance data of similar buildings with
Thermal energy storage in residential buildings: A study of the benefits and impacts
129
similar occupancy and operational conditions. For example, the MKEP energy consumption study analysed by Summerfield et al. (2010) presented a measured average gas based heating energy consumption, for 36 dwellings of an average size of 98m2 floor area and of various built forms 67kWh/day as discussed earlier in Section 2.5.4. The dwellings had energy efficiency measures that corresponded with something close to the 2002 Building Regulations. This published average energy consumption was compared with the average energy demand prediction of around 68.6kWh/day generated by the TRNSYS models with 2002s thermal properties in this research, as detailed in Results Chapter 4, Section 4.5.3. The difference between these figures is approximately 2.4%. The buildings monitored by Summerfield et al. (2010) when compared with those with similar physical, thermal and occupancy conditions as used in this research, indicated that the models represent the real world reasonably well, and that the results generated are realistic.
As discussed previously in Section 2.5.4, Buswell et al. (2013) reported some results from a whole house monitoring trial in the East Midlands, England, where the heat use is disaggregated from high resolution measurements of gas, hot water and the power consumption. The heat and DHW energy consumption measured for a semi- detached and a detached dwelling monitored during the month of January are stated as 73.3kWh/day (58.6kWh/day for space heating and 14.7kWh/day for DHW) and 69.5kWh/day (56.5kWh/day for space heating and 13.0kWh/day for DHW) respectively. Both dwellings had approximately 100m2 floor areas and a thermal insulation roughly corresponding to the 2002 building regulation. They were occupied by 4 adults and 2 children, and 1 adult and 1 child respectively. There is good agreement between the energy consumption they measured for the semi-detached dwelling and the value extrapolated from the simulated mean of 60 day results (approximately 70kWh) of similar dwelling as described in Results Chapter 4 - Section 4.5.3. The difference between the two values is about 3.3kWh (4.5%). Comparing the energy consumption values for space heating only shows a difference of about 3%.
Conversely, there is a relatively large difference, around 9%, between the measured (69.5kWh measured by Buswell et al. (2013)) and the simulated energy consumption (76kWh simulated as described in Results Chapter 4 - Section 4.5.3) for the detached dwelling. The difference is greater when the space heating only figures are
Thermal energy storage in residential buildings: A study of the benefits and impacts
130
compared and is about 15%. One possible explanation for this is that the measurements were taken on a relatively warm day when less heat was needed. The fact that a higher energy consumption was measured for the semi-detached in comparison to that of the detached building, contrary to the generally accepted view that detached dwellings are less energy efficient, can be said to support this assumption. Despite the large difference, it is still well within the minimum and maximum energy consumption variation usually seen, and also simulated in during this work, due to the ambient temperature variation expected over a winter period, which is 60 days from 2nd January in this work.
In summary, the heat demand prediction generated by the models used in this work and the measured data published by Buswell et al. (2013) and Summerfield et al. (2010) as discussed above are in relatively good agreement. Considering the similar size and thermal property parameters used in the models compared to those of the