- Email: [email protected]
Thermal performance of shelter modelling: Improvement of temporary structures
Energy and Buildings 89 (2015) 170–182
Contents lists available at ScienceDirect
Energy and Buildings journal homepage: www.elsevier.com/locate/enbuild
Thermal performance of shelter modelling: Improvement of temporary structures S. Obyn a,∗ , G. van Moeseke a,∗ , V. Virgo b a Architecture et Climat, Faculté d’architecture, d’ingénierie architecturale, d’urbanisme (LOCI), Université Catholique de Louvain, 1 place du Levant, 1348 Louvain-la-Neuve, Belgium b Shelter Research Unit, International Federation of Red Cross Crescent Societies (IFRC), Luxembourg
a r t i c l e
i n f o
Article history: Received 6 October 2014 Received in revised form 4 December 2014 Accepted 18 December 2014 Available online 27 December 2014 Keywords: Simulation Modelling Thermal performance Multi-layer Light transmission Calibration
a b s t r a c t In emergency situations, it is important to provide shelters to protect the population and the support against their environment and to give them some privacy. Unfortunately, tents commonly used in humanitarian context do not ensure comfortable conditions for the occupants. Furthermore, given the very large scale of emergency camps, the intake of fuel in winter condition turns into a major logistical challenge. It is crucial to improve the thermal performance of emergency shelters to (1) increase their indoor comfort and (2) reduce their fuel/wood consumption and related pressure on natural resources. The purpose of this paper is to discuss the difficulties in achieving a realistic thermal model of lightweight structures, taking into account the air permeability of fabrics, their light transmission and the imbrication of several elements (multi-layered shelter). Such a model of the IFRC/ICRC/UNHCR standard family tent is created, based on the building oriented Energy+ thermal simulation model. This model is calibrated and validated by comparing simulation results with in situ measurements realised in the BBRI facility in Belgium, in Burkina Faso and in Luxembourg. This model provides objective assessment of the performance of that shelter for any given context and climate exposure except for night overcooling phenomenon. © 2014 Elsevier B.V. All rights reserved.
1. Introduction There exist many disasters, whether natural or caused by human factors. In emergency situations, it is important to provide shelters to protect the population and the support against their environment and to give them some privacy [1,2]. Unfortunately, tents commonly used in humanitarian context do not ensure comfortable conditions for the occupants, as showed by measurement realised in Burundi from 20 to 25 July 2013 (Fig. 1). Furthermore, such intervention leads to very large-scale emergency camps. Indeed, for example, 117,000 shelters have been constructed between 2002 and end of 2004 and 21,500 additional shelters were planned for 2005 in support to refugees in Afghanistan [3]. In winter conditions, the intake of fuel turns into a major logistical challenge. Scientific researches about emergency sheltering focused on design proposals such as deployable structures [5–9] or adaptable, versatile and compatible construction systems and shelter
∗ Corresponding authors. Tel.: +32 10 47 21 42; fax: +32 10 47 21 50. E-mail addresses: [email protected] (S. Obyn), [email protected] (G. van Moeseke). http://dx.doi.org/10.1016/j.enbuild.2014.12.035 0378-7788/© 2014 Elsevier B.V. All rights reserved.
kits [10–14]. Dedicated design approaches have been proposed for these specific structures, such as 4-dimensional design [11,15]. Very few publications focused on the energy efficiency of these shelters [9,13,16–19]. Crawford et al. [16] proposed a thermal model adapted to a specific shelter. They calibrate the crack area in order to obtain an inside calculated temperature almost equal to the air temperature measured during testing. However, they identified divergences between simulation results obtained with their thermal model and the experimental data, mainly in terms of air change rates. Also, they studied a rigid insulated shelter which behaves more like a standard construction than lightweight structures, as considered in the present research. Consequently, the feasibility of using building oriented simulation model to study shelters has yet to be proved. This lack of interest on the thermal behaviour of emergency shelters is surprising since inappropriate hygro-thermal conditions outdoors and indoors have proven to be related to various health hazard and stress, both in cold and warm conditions [20–22]. Our contribution aims three objectives. Firstly, this contribution proposes a framework for further studies about such shelters. Secondly, based on field measurements, it documents the thermal behaviour of the most popular family shelter among emergency
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
171
important to ensure a proper interpretation of the results. Finally, it might be useful to compare the performance of alternative designs proposed by various authors. This will be possible only if all studies consider similar conditions. For all these reasons, we propose in the next section standard hypotheses, and weather data to be used for thermal model comparison of shelters. 2. Hypothesis 2.1. Shelter description Fig. 1. Measurements in Burundi [4].
agencies under various climates. This documentation intends awareness rising of the poor thermal behaviour and comfort conditions of such shelters. Thirdly, it discusses the difficulties to establish a thermal model of such shelter and shows the accuracy we achieved while trying to create one starting from traditional building physic tools. This discussion will help further developments of dedicated models. A building physic analysis of family shelters or other emergency housing should aim two main objectives: (1) improving indoor conditions through passive design, both for cold and hot climates, in order not to threaten the health of disasters victims; (2) reducing fuel demands for heating in order to make fuel supply easier and less expensive. To do so, alternative designs for shelters may be proposed. But these alternatives have to consider their impact on the other specificities of the shelter in a holistic way. Key parameters to consider are [23]: 1. the geometry: warehouse, transport and handling aspects limit the weight and size of the shelter; 2. the social aspect: size, intimacy and adaptability of the shelter must be considered; 3. other constructive aspects such as mechanical resistance. Anyhow, it is obvious no single discipline could develop an ideal shelter by itself. A multidisciplinary approach must be preferred. About building physics, field experiments are obviously the most accurate way to collect data and compare existing alternative design and actual conditions. Thermal models should be used to test new designs before the prototypal stage. The relevance of such models depends on their ability to reproduce the actual behaviour of a tent. Therefore it is useful to start by assessing the accuracy of the developed thermal model by confronting it to monitoring values. It might also be useful to compare the accuracy of the model with the one achieved by previous models. Such information is
The studied tent is known as a family shelter for emergency situations. It is used by the UNHCR, the IFRC and the ICRC agencies [24]. This tent is composed of one outer skin and one inner skin. The inner skin envelops a central space that may be divided in two equal parts. The outer skin creates two transitional spaces in front of both entrances. Figs. 2–4 show plans and views of that tent. A 10 cm gap exists laterally between the inner and the outer skins. Openings are set on both sides of the central space and on outer skin in transitional spaces. A moveable flap and a mosquito net close them. Ventilation openings exist above both doors. Table 1 summarises the main characteristics of the materials, based on laboratory measurements. Composition, weight, solar and visual transmittance and reflexions and air and water permeability’s are measurement results made on samples from an actual tent. Thermal conductivities are indicative values based on literacy [25]. 2.2. Field measurements Monitoring’s were conducted on three different climates: Brussels in Belgium, Sag Nioniogo in Burkina Faso and Bertrange in Luxembourg. Fig. 15 shows temperatures and radiation values observed during field experiments for the first period of measurement in Brussels. Investigations in the Brussels area took place at the BBRI research facility of Limelette during the months of October and November 2013 (Figs. 5 and 6). It was a true experimental investigation, out of emergency situation. The shelter was monitored thanks to 13 thermocouples (Fig. 6) for a period of 39 days in total distributed in three periods: from 4 to 10 October 2013 with all openings closed, from 10 to 24 October and from 7 to 26 November 2013 with inner and outer vents opened. Extensive meteorological data are available from an on-site weather station. Fig. 6 describes the position of data acquisition material used to collect these data in Brussels. The thermocouples are placed in order to identify the thermal stratification inside the tent and the thermal behaviour of the gap between the inner and outer tent when exposed to the sun. For the acquisition, a computer is placed in the inner tent and provides 12.18 W/m2 of internal gain.
Fig. 2. View and photo of the studied tent [24].
172
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Fig. 3. Views and elevations of the tent [24].
Table 1 Materials properties. Inner skin
Outer wall
Outer roof
Mud flaps and ground sheet
Composition
Poly-cotton 140 1.2 27.50 47.20 20.20 39.20 – –
49.7% Cotton 50.3% Polyester 373 0.2 8.52 64.48 7.66 76.48 3.88 1410
HDPE fabric + PE coating
Weight [g/m2 ] ISO 3801-5 Thermal conductivity [W/mK] Heat transmittance [%] EN 410 Heat reflection [%] EN 410 Light transmittance [%] EN 410 Light reflection [%] EN 410 Air permeability @ 100 Pa [l/(m2 s)] ISO 9237 Water vapour permeability [g/(24 h m2 )] – ASTM E96
46.4% Cotton 53.6% Polyester 239 0.2 13.38 62.15 12.74 73.76 30.10 1460
165 0.5 0.46 34.64 0.09 43.54 <0.10 20
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
173
Fig. 4. Plan of the tent [24].
Investigations in Burkina Faso were conducted between 6 November and 2 December 2013 during an actual emergency crisis. However, the monitored tent was not occupied and all openings stayed closed during monitoring period. Field data are more
limited due to the inherent difficulty of such theatre of investigations. Internal and external temperatures are measured thanks to 2 sensors. Irradiation data were not available but can be rebuild thanks to geographic and time interpolation between typical years. Indeed the field experiments were conducted in a sunny period. Daily solar irradiation profiles then only depends of the sun position. In the case of field measurement in Luxembourg, the indoor temperature of the shelter was monitored thank to 1 sensor. The outdoor temperature as well as the wind’s speed and direction were not measured on the experiment site but considered data correspond to the airport localised at approximately 13 km at North-East of the experimentation place. Note also that no solar radiation data is available excepted 30 years daily mean value of radiation and number of minutes of insulation per hour at the airport of Luxembourg. 2.3. Thermal model: hypotheses
Fig. 5. Site of measurements in BBRI research facility, Belgium.
Fig. 6. Position of thermocouples for investigations in Brussels.
Many thermal models exist to analyse buildings. Energy+ and Trnsys are often chosen when it comes to academic research [26]. In this study, Energy+ along with the Sketchup Energy+ plugin are used in order to create a 3 dimensional model with an accurate evaluation of view factor for radiative heat exchange between surfaces. A shelter shows specific properties for which the use of building thermal model software is not obvious. Those properties are related to the skin fabric that is neither airtight nor opaque. Other specificities are the very low insulation level, the very small thermal mass and the major influence of the ground temperature. Also, the interaction between the inner and outer skin has to be considered. The model we develop deals with these specificities by dividing the shelter in seven thermal zones. Both a resistive model and an airflow model connect these zones (Fig. 7). This way, the gap between the inner and outer tents is modelled in a similar way as a double skin fac¸ade, and the model may be compared to the data collected during the field measurement. All adjacent surfaces between zones, or between zones and the outside, are considered as
174
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Fig. 7. Air model.
fully glazed. The glazing photometric properties are defined based in Table 1. This way, solar transmission through the fabric is considered. Only the ground sheet is modelled as an opaque material. The thermal conductivity of the glazing is also adapted to match values specific to the various fabrics used. The airtightness of the tent is the main difficulty. Indeed the fabrics are not airtight. Further, windows and doors are closed by Velcro bands only. The connection to the ground also shows air leakages. Fig. 7 details the air model we created to deal with such behaviour. Airflows through the outer tent are related to both the temperature difference and the wind pressure. The
effect of the temperature difference is easily modelled by the Archimedes equation. The wind pressure is modelled by using pressure coefficients which are determined thanks to TNO Cp Generator [27] (see Appendix). Note that the main obstacles around the tent in the site of measurements were defined to obtain the most accurate values of wind pressure coefficient on the tent. The values we evaluated for the test case in Limelette are presented in Appendix. All openings are described by an area, a flow exponent and a flow coefficient. An indicative area for all openings is determined based on the actual geometry of the tent. To consider the air infiltration throw cracks and fabric, all glazed surface are considered as fully
Fig. 8. 4–10 October calibration: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
opened and the discharge coefficient of each glazed surface is calculated to correspond to the air permeability of the corresponding fabric and crack. Flow exponents are related to the king of opening (crack or large opening) and indicative values are easily found in the thermal software documentation. In this study, only large openings are modelled, with a flow exponent of 0.65, to represent air infiltration as well throw fabrics as throw crack. This way, the only uncertainty is about the discharge coefficient of the openings. 3. Model calibration The thermal model of the emergency family shelter is calibrated using the field measurements realised in the BBRI facility in Brussels, Belgium. The parameters to calibrate are the discharge coefficients used to model the air infiltration throw cracks and fabric of the tent and the soil thickness interacting in thermal behaviour of the shelter. To determine these coefficients, we proceed this way: firstly we simplify the model by considering that only five different values of discharge coefficient are to be identified, i.e. two for the outer tent openings (windows and vents separately), one for the mud flaps (mainly to represent crack infiltration), one for the inner tent fabric and one for the inner vents – considering only air permeability of the fabric for other surfaces (outer roof and walls), based on laboratory measurements. We evaluated also some soil thickness under the ground sheet which may be consider as part of the composition of the floor of the tent,
175
distinguishing lateral zones, transition zones and inner zone. Secondly, we make different simulations for various values of flow coefficients and soil thickness and calculate the mean variation between the simulation results and monitored temperatures for the monitored zones and especially the zone describing the inner tent. This exercise is done for the three monitoring periods in Brussels (i.e. with all openings closed and with external and internal vents opened). Some results for the first period of measurements (i.e. with all openings closed) are shown in Fig. 8. Based on all results, the difference between the calculated and the measured indoor temperatures is characterised by two polynomial equations of the third degree calculated by the statistical program JMP [28]; i.e. one for all openings closed and one for the tent with external and internal vents opened. The values of discharge coefficients and soil thickness leading to the minimum indoor temperature differences between measurements and calculations are identified by resolving the polynomial equations in excel, giving limit values of each parameter. Results are presented in Table 2. 4. Results and discussion 4.1. Quality of the model calibration Fig. 8 presents the results of calibrated tent model for the first periods of measurements in Belgium, with all opening closed. Fig. 8a presents the measured and calculated indoor temperatures
Fig. 9. 10–24 October calibration: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
176
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Fig. 10. 7–27 November calibration: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
and the outdoor temperature. Fig. 8b illustrates the hourly indoor temperature difference between calculation and measurement (sticks) and the confidence interval of this temperature difference on the considered investigation period so that the temperature difference ranges in this interval during 95% of the time. The correspondent values of discharge coefficient and soil thickness are presented in Table 2. These lead to a mean deviation between measured and calculated temperature of 1.94 ◦ C. We observe that the value of discharge coefficient of outer vents have a small impact on goodness of the model. Considering only daily period, the best ground temperature to consider is 2 ◦ C higher than for the whole period. In the model, we can only choose a constant ground temperature or a ground temperature equal to outdoor temperature. The ground temperature is thus difficult to model but
Table 2 Calibration results. Parameter
Calibrated value
Discharge coefficients Inner tent Mud flaps External windows, closed External vents, closed Internal vents, closed External vents, opened Internal vents, opened
0.0242 0.0052 0.0840 0.1595 0.1595 0.1595 0.2300
Soil thickness participating to floor composition of Transitional and inner zones Lateral zones
15 cm 5 cm
Table 3 Simulation results for calibration field measurements periods. Period of measurement
Measured temperature
[◦ C]
Calculated temperature
[◦ C]
Difference between measurement and calculation
4–10 October
Maximum Minimum Variation
31.8 6.3 25.5
Maximum Minimum Variation
28.8 2.3 26.5
3.0 ◦ C 4.0 ◦ C 1.0 ◦ C
10–24 October
Maximum Minimum Variation
29.8 2.7 27.1
Maximum Minimum Variation
28.8 2.3 26.5
1.0 ◦ C 0.4 ◦ C 0.6 ◦ C
7–27 November
Maximum Minimum Variation
23.2 1.0 22.2
Maximum Minimum Variation
28.0 1.5 26.5
4.8 ◦ C 0.5 ◦ C 4.3 ◦ C
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182 Table 4 Ground temperatures considered for simulations in Burkina Faso. Constant value of ground temperature at
[◦ C]
Building surface Deep Shallow
30 28 29
has a large impact on result, i.e. in order of +0.12 ◦ C in mean deviation per degree in ground temperature. We may also note that it is difficult to reproduce the night overcooling in case of clear sky. During the periods of 10–24 October and 7–11 November, the inner and the outer vents are opened and their discharge coefficients are presented in Table 2. The mean deviations between
177
measured and calculated internal temperatures are respectively 1.59 ◦ C and 1.77 ◦ C. Results are presented in Figs. 9 and 10. Finally, a summary of the simulation results is presented in Table 3. The model represents between 98% and 119% of the internal temperature variation during the three periods of calibration considered. 4.2. Validation of model calibration in summer conditions in Burkina Faso We may then check the calibration of the thermal model with measurements realised in Burkina Faso for the Family Tent with all openings closed. Simulation results with the calibrated model
Fig. 11. Field measurement in Burkina Faso [30].
Fig. 12. Some simulation results in Burkina Faso [30]: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
178
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Table 5 Minimum and maximum internal temperatures: measured versus calculated. Measured temperature
[◦ C]
Calculated temperature
[◦ C]
Difference between measurement and calculation
Maximum Minimum Variation
51.4 18.9 32.5
Maximum Minimum Variation
49.0 20.5 28.5
2.4 ◦ C 1.6 ◦ C 4.0 ◦ C
Table 6 Minimum and maximum internal temperatures: measured versus calculated. Measured temperature
[◦ C]
Calculated temperature
[◦ C]
Difference between measurement and calculation
Maximum Minimum Variation
28.8 −3.7 32.5
Maximum Minimum Variation
31.1 −1.1 32.2
2.3 ◦ C 2.6 ◦ C 0.3 ◦ C
are presented in Fig. 12. The in situ measurements concern the outdoor temperature and the indoor temperature of the Family Tent. No information about wind speed, ground temperature and solar radiation during the field measurement is available. Nevertheless, the weather was sunny during this period. It is thus possible to reconstruct the solar radiation based on the location of the considered shelter. The weather data considered for the simulation are based on the climate file for Ouagadougou available on EnergyPlus website [29], except for the dry bulb outdoor temperature
which is modified to correspond to field measurements. The ground temperature is defined as presented in Table 4. The construction method of the meteorological input file considered for the calculation of the internal temperature of the shelter explains mainly the differences observed between measurement and calculation. Indeed, the difference between calculated and measured internal temperature in the Family Tent (Fig. 12) is the most important from 11 to 13 November 2013. During these days, the wind speed in the climate file considered is very high. Maybe
Fig. 13. Simulation results in Luxembourg [31]: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
179
Fig. 14. Simulation result in Brussels.
the in situ wind speed is lower than the considered wind speed for the calculation. The mean internal temperature difference between calculation and measurement for the Family Tent is 2.46 ◦ C. On the other hand, the maximum and minimum measured internal temperatures during the field measurements are respectively 51.4 ◦ C and 18.9 ◦ C and it appears respectively on 13 and 25 November. The calculated maximum and minimum internal temperatures during the field measurements are respectively 49.0 ◦ C and 20.5 ◦ C and it appears on 18 November and 2 December. The model may thus represent 87.7% of the internal temperature variation, as presented in Table 5 (Fig. 11).
The weather was not fully sunny then the solar radiation must be reconstruct on base of time of insulation and daily mean value of solar radiation for 30 years. The weather data are then not accurate and the simulation results are consequently not very good. Nevertheless, it appears that the most difficult thing to reproduce is the night overcooling in case of clear sky, as presented in Fig. 13. However, the model reproduces 99% of the internal temperature variation over the measurement period, which is pretty good, as presented in Table 6.
4.4. Solar radiation and wind speed 4.3. Validation of model calibration in winter conditions in Luxembourg [31] The field measurement in Luxembourg is difficult to reproduce due to the lack of meteorological information. Indeed, the outdoor temperature as well as the wind’s speed and direction available correspond to the airport localised at approximately 13 km at North-East of the experimentation place. Also, no solar radiation data is available excepted 30 years daily mean value of radiation and number of minutes of insulation per hour at the airport of Luxembourg.
Fig. 14 presents the complete meteorological data for the first period of measurement in Brussels, Belgium. It appears a large dependence of the indoor temperature to as well solar radiation then wind speed. The thermal model reproduced this dependence with a pretty good reproduction of thermal behaviour of the shelter during daily periods. However, the difficulty to represent night overcooling in the case of clear sky may be seen during nights from 6 to 7 October and from 9 to 10 October.
Table 7 Points of wind pressure coefficients calculation.
Fig. 15. Site of measurements in Belgium.
Calculation point
X [cm]
Y [cm]
Z [cm]
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
0 65 235 595 660 595 235 65 65 65 235 595 595 595 235 65
200 65 0 65 200 335 400 335 200 100 100 100 200 300 300 300
95 105 100 105 95 105 100 105 135 135 135 135 135 135 135 135
180
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Fig. 16. Projections for Cp calculation in Belgium.
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
181
Table 8 Cp values at each point of calculation for different wind directions. Wind direction Calculation point
0 North
45
90 East
135
180 South
225
270 West
315
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
−0.688 −0.258 −0.140 −0.258 −0.697 0.048 0.280 0.044 0.220 −0.280 −0.130 −0.030 0.244 0.036 0.214 0.034
0.037 −0.698 −0.234 −0.138 −0.258 −0.586 0.061 0.271 0.103 −0.940 −0.188 −0.740 0.103 −0.448 0.089 0.200
0.262 0.054 −0.572 −0.239 −0.133 −0.239 −0.572 0.054 −0.738 0.070 −0.391 −0.270 −0.738 −0.204 −0.859 0.080
0.037 0.271 0.061 −0.586 −0.258 −0.138 −0.234 −0.698 −0.257 −0.700 0.039 −0.390 −0.265 −0.130 −0.267 −0.940
−0.688 0.044 0.280 0.044 −0.688 −0.258 −0.140 −0.258 −0.127 −0.250 0.214 0.070 −0.133 −0.280 −0.130 −0.280
−0.248 −0.569 0.056 0.259 0.039 −0.698 −0.228 −0.138 −0.154 −0.400 0.082 0.230 −0.167 −0.939 −0.188 −0.300
−0.133 −0.239 −0.572 0.054 0.262 0.054 −0.572 −0.239 −0.301 −0.200 −0.859 0.080 −0.331 0.075 −0.391 −0.270
−0.261 −0.153 −0.233 −0.706 0.042 0.166 0.067 −0.586 0.045 −0.140 −0.268 0.240 0.051 0.170 0.039 −0.450
However, some phenomena are difficult to reproduce, as overcooling during night in clear sky conditions. Also, the ground temperature has a wide impact on results but is difficult to model. To conclude, using windows to model fabric of tent permits, by adapting discharge coefficients and photometric properties, to reproduce the properties of air permeability and luminous transmittance of fabrics and obtain a good representation of light structure thermal behaviour. Other similar models may be constructed in view to compare form, material, and design to assess the quality of such installation. Therefore, the methodology described in Section 2.3 is a methodological recommendation for further studies.
Acknowledgement Fig. 17. Positions of wind pressure coefficient calculation.
5. Conclusion The aim of this paper is to discuss the creation of a model representing the thermal behaviour of the family shelter for emergency situations used by the UNHCR, the IFRC and ICRC [24]. The main hypothesis (described in Section 2.3) considered for the model construction are: (1) the model is divided in several zones in order to represent the thermal behaviour of the gap between inner and outer skin; (2) these zones are connected with both a resistive and an airflow model; (3) all fabric surfaces (except ground sheet) are considered as fully glazed to represent the translucence of the fabrics; (4) all these glazed surfaces are considered as opened with an adapted discharge coefficient to represent the airtightness of the shelter. After calibration with field experiment data for Belgium, the model is relevant and may reproduce the actual behaviour of a tent, as well for Belgium than for other climatic conditions (as shown for Burkina Faso and Luxembourg). The deviation calculated between the measured and calculated internal temperature in Brussels, for different conditions (i.e. vents opened or closed), ranges between 1.6 and 1.9 ◦ C for the three periods of measurement considered. When the calibrated model is transposed in Burkina Faso, the value of the mean deviation between measured and calculated indoor temperature is 2.5 ◦ C and the model reproduces 87.7% of the internal temperature variation (Tmax –Tmin ) within the considered period. Respectively, when considering winter condition as evaluated in Luxembourg, the mean deviation over the field measurement period is around 2.5 ◦ C and the model reproduces 99% of the internal temperature variation.
This research was partly conducted in the frame of the SIMBA FEDER project, supported by the Walloon Region and the European Union.
Appendix A. Appendix A.1. Cp values in Limelette The values of wind pressure coefficients are determined thanks to TNO Cp Generator [27]. The values corresponding to the test case in Limelette are presented in Table 8. Note that the main obstacles around the tent in the site of measurements were defined to obtain the most accurate values of wind pressure coefficient on the tent. The assumptions taken for the evaluation of Cp values with the Cp Generator are as follows: - The environment is defined as cultivated land, that is to say a plain with small obstacles (the obstacle distance is greater than twenty times the height of the obstacle): z0 = 0.5 m; - The weather station of the BBRI was taken as reference (about 30 m from the shelter) with a height of the wind speed measurement of 12 m; - The grove, the tree row and the house are modelled as a major obstacles (Fig. 15). With these assumptions, the height considered to calculate the Cp values is the height of wind speed measurement (12 m) and the local roughness at the measuring point is regarded as identical to the surrounding terrain. This allows to calculate the wind pressure directly from local wind data without correction. In other words, the Cp values are corrected for the data of the local weather station.
182
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Note also that only rectangular building may be introduced in the TNO Cp Generator. The trick is then to make rectangular projections of the tent that approach the blockage of the wind. There are three typical facades: - Long side; - Short side; - Oblique side. The other five sides may be derived from cyclic substitution. Hence, three different projections are realised, as shown in Fig. 16. The results of Cp value on each fac¸ade (Fig. 17) are given in Tables 7 and 8. References [1] J. Ashmore, Tents: A Guide to the Use and Logistics of Family Tents in Humanitarian Relief, United Nations Publication, 2004, OCHA Ref Nr. OCHA/ESB/2004/19. http://josephashmore.org/publications/tents.pdf. [2] UNHCR, Handbook for Emergencies, United Nations High Commissioner for Refugees, Geneva, 2000, pp. 405. [3] P. Foley, UNHCR – Shelter Programme Monitoring and Evaluation: Final Report, 2005, pp. 6. [4] IFRC – SRU Field Tests & Findings Research and Development Project Launched by IFRC IRC UNHCR on Tents Investigations, Lux Red Cross, Bujumbura, Burundi, July–Augustus 2013. [5] L. Alegria Mira, A.P. Thrall, N. De Temmerman, Deployable scissor arch for transitional shelters, Autom. Constr. 43 (2014) 123–131. [6] P.J. Owens, A. Forgione Jr., S. Briggs, Challenges of international disaster relief: use of a Deployable Rapid Assembly Shelter and Surgical Hospital, Disaster Manag. Response 3 (1) (2005) 11–16. [7] P. Manfield, A comparative study of temporary shelters used in cold climates. What can be learnt from the design of the Yurt and the Scott tent to inform the future design of shelters systems for emergency relief? in: History Essay for the MPhil Degree in Environmental Design in Architecture, Cambridge University, Cambridge, UK, 2000 http://www.shelterproject.org/ downloads/manfield1.pdf [8] P. Manfield, Modelling of a Cold Climate Emergency Shelter Prototype and a Comparison with the United Nations Winter Tent. Technical Essay for the MPhil Degree in Environmental Design in Architecture, Cambridge University, Cambridge, UK, 2000 http://shelterproject.org/downloads/manfield2.pdf [9] M. Asefi, F. Ahangar Sirus, Transformable shelter: evaluation and new architectural design proposals Procedia Soc. Behav. Sci. 51 (2012) 961–966. [10] J. Roekens, V. Virgo, L. De Leat, M. Van Craenenbroeck, S. Puystiens, M. Mollaert, A new type of sheltering for disaster affected sites: the ‘Clever roof’, in: Proceedings of the International Association for Shell and Spatial Structures (IASS) Symposium 2013, Wroclaw University of Technology, Poland, 2013. [11] C. Henrotay, A contribution to an integrated and more sustainable design approach for the material support of shelter after a disaster (Ph.D. dissertation), Vrije Universiteit Brussel, Brussels, 2008. [12] R. Battilana, Design of Cold Climate Temporary Shelter for Refugees, 2000 http://www.shelterproject.org/downloads/cold%20climate%20liner.pdf
[13] C. Hale, R. Dooley, Keeping People Warm: Collaborative R&D, ShelterBox, 2014 http://www.shelterbox.org/uploads/Sources/Downloads/shelterbox trust annual report 2012.pdf [14] C. Henrotay, M. Mollaert, H. Hendrickx, Adaptable Construction Systems for Shelter in Emergency Situation. Adaptables 2006 Proceedings of the Joint CIB, Tensinet, IASS International Conference on Adaptability in Design and Construction, vol. 2, No. 4, 2006, pp. 63–67. [15] C. Henrotay, W. Debacker, M. Mollaert, W. De Wilde, H. Hendrickx, 4Dimensional Design: A Design Strategy for Efficient Shelter and Sustainable Housing After Conflict-based and Natural Disasters. Management of Natural Resources, Sustainable Development and Ecological Hazards, vol. 1, WIT Press, 2006, pp. 341–350. [16] C. Crawford, P. Manfield, A. McRobie, Assessing the thermal performance of an emergency shelter system, Energy Build. 37 (5) (2005) 471–483. [17] X.W. Winston, R.E. England, K.C. Kuivinen, J.J. Potter, N.S. Krug, A critical review of design and use of field tent shelter in polar regions, Polar Rec. 34 (189) (1998) 113–122. [18] P. Manfield, T. Corsellis, Cold Climate Emergency Shelter Systems. A Research Project for Humanitarian Organisations, Unpublished Paper for the Halley Stewart Trust, Cambridge, UK, 1999 http://www.shelterproject.org/ downloads/coldshelter.pdf [19] P. Manfield, Emergency Shelter for Humanitarian Relief in Cold Climates: Policy and Praxis, Unpublished Dissertation for a Diploma in Architecture, Martin Centre for Architectural and Urban Studies, Department of Architecture, Cambridge University, Cambridge, UK, 2001 http://www.shelterproject.org/ downloads/coldshelter2.pdf [20] J. Rudge, R. Gilchrist, Measuring the health impact of temperatures in dwellings: investigating excess winter morbidity and cold homes in the London Borough of Newham, Energy Build. 39 (7) (2007) 847–858. [21] H. Johnson, R.S. Kovats, G. McGregor, J. Stedman, M. Gibbs, H. Walton, The impact of the 2003 heat wave on daily mortality in England and Wales and the use of rapid weekly mortality estimates, Eurosurveillance 10 (7–9) (2005) 68–171. [22] J. Krieger, D.L. Higgins, Housing and health time again for public health action. Public health matters, Am. J. Public Health 92 (May (5)) (2002). [23] J. Damerell, T. van Zutphen, The Sphere Project: Humanitarian Charter and Minimum Standards. Humanitarian Response, Edition, 2011, ISBN: 978-1908176-004, http://www.ifrc.org/docs/idrl/I1027EN.pdf [24] UNHCR (United Nations High Commissioner for Refugees), Family Tent: UNHCR Item No. 05353, 2011. [25] DesignBuilder (Version 3.2.0.073) © Copyright 200-2013, DesignBuilder Software Ltd. [26] D.B. Crawley, J.W. Hand, M. Kummert, B.T. Griffith, Contrasting the capabilities of building energy performance simulation programs, Build. Environ. 43 (2008) 661–673. [27] Nederlandse Organisatie voor Toegepast Natuurwetenschappelijk Onderzoek (TNO). TNO Webapplications, http://cpgen.tno.nl/cp/info.asp [28] JMP® 10.0.0 Edition en 64 bits, Copyright © 2012 SAS Institute Inc. All Rights Reserved. [29] U.S. Department of Energy Website. http://apps1.eere.energy.gov/buildings/ energyplus/weatherdata request search.cfm?sortKey=country&opt=1, accesses 01.08.14. [30] IFRC–SRU Field Tests Investigations Tough Speedkits Project, Comparison Between Standard Family Tent With and Without Shade Net, Sag Nioniogo, Burkina Faso, November 2013. [31] IFRC–SRU Field Tests & Findings, Research on Winterization Potential Solutions, IFRC–SRU Field Tests & Findings, Bertrange, Luxembourg, March 2014.
Contents lists available at ScienceDirect
Energy and Buildings journal homepage: www.elsevier.com/locate/enbuild
Thermal performance of shelter modelling: Improvement of temporary structures S. Obyn a,∗ , G. van Moeseke a,∗ , V. Virgo b a Architecture et Climat, Faculté d’architecture, d’ingénierie architecturale, d’urbanisme (LOCI), Université Catholique de Louvain, 1 place du Levant, 1348 Louvain-la-Neuve, Belgium b Shelter Research Unit, International Federation of Red Cross Crescent Societies (IFRC), Luxembourg
a r t i c l e
i n f o
Article history: Received 6 October 2014 Received in revised form 4 December 2014 Accepted 18 December 2014 Available online 27 December 2014 Keywords: Simulation Modelling Thermal performance Multi-layer Light transmission Calibration
a b s t r a c t In emergency situations, it is important to provide shelters to protect the population and the support against their environment and to give them some privacy. Unfortunately, tents commonly used in humanitarian context do not ensure comfortable conditions for the occupants. Furthermore, given the very large scale of emergency camps, the intake of fuel in winter condition turns into a major logistical challenge. It is crucial to improve the thermal performance of emergency shelters to (1) increase their indoor comfort and (2) reduce their fuel/wood consumption and related pressure on natural resources. The purpose of this paper is to discuss the difficulties in achieving a realistic thermal model of lightweight structures, taking into account the air permeability of fabrics, their light transmission and the imbrication of several elements (multi-layered shelter). Such a model of the IFRC/ICRC/UNHCR standard family tent is created, based on the building oriented Energy+ thermal simulation model. This model is calibrated and validated by comparing simulation results with in situ measurements realised in the BBRI facility in Belgium, in Burkina Faso and in Luxembourg. This model provides objective assessment of the performance of that shelter for any given context and climate exposure except for night overcooling phenomenon. © 2014 Elsevier B.V. All rights reserved.
1. Introduction There exist many disasters, whether natural or caused by human factors. In emergency situations, it is important to provide shelters to protect the population and the support against their environment and to give them some privacy [1,2]. Unfortunately, tents commonly used in humanitarian context do not ensure comfortable conditions for the occupants, as showed by measurement realised in Burundi from 20 to 25 July 2013 (Fig. 1). Furthermore, such intervention leads to very large-scale emergency camps. Indeed, for example, 117,000 shelters have been constructed between 2002 and end of 2004 and 21,500 additional shelters were planned for 2005 in support to refugees in Afghanistan [3]. In winter conditions, the intake of fuel turns into a major logistical challenge. Scientific researches about emergency sheltering focused on design proposals such as deployable structures [5–9] or adaptable, versatile and compatible construction systems and shelter
∗ Corresponding authors. Tel.: +32 10 47 21 42; fax: +32 10 47 21 50. E-mail addresses: [email protected] (S. Obyn), [email protected] (G. van Moeseke). http://dx.doi.org/10.1016/j.enbuild.2014.12.035 0378-7788/© 2014 Elsevier B.V. All rights reserved.
kits [10–14]. Dedicated design approaches have been proposed for these specific structures, such as 4-dimensional design [11,15]. Very few publications focused on the energy efficiency of these shelters [9,13,16–19]. Crawford et al. [16] proposed a thermal model adapted to a specific shelter. They calibrate the crack area in order to obtain an inside calculated temperature almost equal to the air temperature measured during testing. However, they identified divergences between simulation results obtained with their thermal model and the experimental data, mainly in terms of air change rates. Also, they studied a rigid insulated shelter which behaves more like a standard construction than lightweight structures, as considered in the present research. Consequently, the feasibility of using building oriented simulation model to study shelters has yet to be proved. This lack of interest on the thermal behaviour of emergency shelters is surprising since inappropriate hygro-thermal conditions outdoors and indoors have proven to be related to various health hazard and stress, both in cold and warm conditions [20–22]. Our contribution aims three objectives. Firstly, this contribution proposes a framework for further studies about such shelters. Secondly, based on field measurements, it documents the thermal behaviour of the most popular family shelter among emergency
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
171
important to ensure a proper interpretation of the results. Finally, it might be useful to compare the performance of alternative designs proposed by various authors. This will be possible only if all studies consider similar conditions. For all these reasons, we propose in the next section standard hypotheses, and weather data to be used for thermal model comparison of shelters. 2. Hypothesis 2.1. Shelter description Fig. 1. Measurements in Burundi [4].
agencies under various climates. This documentation intends awareness rising of the poor thermal behaviour and comfort conditions of such shelters. Thirdly, it discusses the difficulties to establish a thermal model of such shelter and shows the accuracy we achieved while trying to create one starting from traditional building physic tools. This discussion will help further developments of dedicated models. A building physic analysis of family shelters or other emergency housing should aim two main objectives: (1) improving indoor conditions through passive design, both for cold and hot climates, in order not to threaten the health of disasters victims; (2) reducing fuel demands for heating in order to make fuel supply easier and less expensive. To do so, alternative designs for shelters may be proposed. But these alternatives have to consider their impact on the other specificities of the shelter in a holistic way. Key parameters to consider are [23]: 1. the geometry: warehouse, transport and handling aspects limit the weight and size of the shelter; 2. the social aspect: size, intimacy and adaptability of the shelter must be considered; 3. other constructive aspects such as mechanical resistance. Anyhow, it is obvious no single discipline could develop an ideal shelter by itself. A multidisciplinary approach must be preferred. About building physics, field experiments are obviously the most accurate way to collect data and compare existing alternative design and actual conditions. Thermal models should be used to test new designs before the prototypal stage. The relevance of such models depends on their ability to reproduce the actual behaviour of a tent. Therefore it is useful to start by assessing the accuracy of the developed thermal model by confronting it to monitoring values. It might also be useful to compare the accuracy of the model with the one achieved by previous models. Such information is
The studied tent is known as a family shelter for emergency situations. It is used by the UNHCR, the IFRC and the ICRC agencies [24]. This tent is composed of one outer skin and one inner skin. The inner skin envelops a central space that may be divided in two equal parts. The outer skin creates two transitional spaces in front of both entrances. Figs. 2–4 show plans and views of that tent. A 10 cm gap exists laterally between the inner and the outer skins. Openings are set on both sides of the central space and on outer skin in transitional spaces. A moveable flap and a mosquito net close them. Ventilation openings exist above both doors. Table 1 summarises the main characteristics of the materials, based on laboratory measurements. Composition, weight, solar and visual transmittance and reflexions and air and water permeability’s are measurement results made on samples from an actual tent. Thermal conductivities are indicative values based on literacy [25]. 2.2. Field measurements Monitoring’s were conducted on three different climates: Brussels in Belgium, Sag Nioniogo in Burkina Faso and Bertrange in Luxembourg. Fig. 15 shows temperatures and radiation values observed during field experiments for the first period of measurement in Brussels. Investigations in the Brussels area took place at the BBRI research facility of Limelette during the months of October and November 2013 (Figs. 5 and 6). It was a true experimental investigation, out of emergency situation. The shelter was monitored thanks to 13 thermocouples (Fig. 6) for a period of 39 days in total distributed in three periods: from 4 to 10 October 2013 with all openings closed, from 10 to 24 October and from 7 to 26 November 2013 with inner and outer vents opened. Extensive meteorological data are available from an on-site weather station. Fig. 6 describes the position of data acquisition material used to collect these data in Brussels. The thermocouples are placed in order to identify the thermal stratification inside the tent and the thermal behaviour of the gap between the inner and outer tent when exposed to the sun. For the acquisition, a computer is placed in the inner tent and provides 12.18 W/m2 of internal gain.
Fig. 2. View and photo of the studied tent [24].
172
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Fig. 3. Views and elevations of the tent [24].
Table 1 Materials properties. Inner skin
Outer wall
Outer roof
Mud flaps and ground sheet
Composition
Poly-cotton 140 1.2 27.50 47.20 20.20 39.20 – –
49.7% Cotton 50.3% Polyester 373 0.2 8.52 64.48 7.66 76.48 3.88 1410
HDPE fabric + PE coating
Weight [g/m2 ] ISO 3801-5 Thermal conductivity [W/mK] Heat transmittance [%] EN 410 Heat reflection [%] EN 410 Light transmittance [%] EN 410 Light reflection [%] EN 410 Air permeability @ 100 Pa [l/(m2 s)] ISO 9237 Water vapour permeability [g/(24 h m2 )] – ASTM E96
46.4% Cotton 53.6% Polyester 239 0.2 13.38 62.15 12.74 73.76 30.10 1460
165 0.5 0.46 34.64 0.09 43.54 <0.10 20
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
173
Fig. 4. Plan of the tent [24].
Investigations in Burkina Faso were conducted between 6 November and 2 December 2013 during an actual emergency crisis. However, the monitored tent was not occupied and all openings stayed closed during monitoring period. Field data are more
limited due to the inherent difficulty of such theatre of investigations. Internal and external temperatures are measured thanks to 2 sensors. Irradiation data were not available but can be rebuild thanks to geographic and time interpolation between typical years. Indeed the field experiments were conducted in a sunny period. Daily solar irradiation profiles then only depends of the sun position. In the case of field measurement in Luxembourg, the indoor temperature of the shelter was monitored thank to 1 sensor. The outdoor temperature as well as the wind’s speed and direction were not measured on the experiment site but considered data correspond to the airport localised at approximately 13 km at North-East of the experimentation place. Note also that no solar radiation data is available excepted 30 years daily mean value of radiation and number of minutes of insulation per hour at the airport of Luxembourg. 2.3. Thermal model: hypotheses
Fig. 5. Site of measurements in BBRI research facility, Belgium.
Fig. 6. Position of thermocouples for investigations in Brussels.
Many thermal models exist to analyse buildings. Energy+ and Trnsys are often chosen when it comes to academic research [26]. In this study, Energy+ along with the Sketchup Energy+ plugin are used in order to create a 3 dimensional model with an accurate evaluation of view factor for radiative heat exchange between surfaces. A shelter shows specific properties for which the use of building thermal model software is not obvious. Those properties are related to the skin fabric that is neither airtight nor opaque. Other specificities are the very low insulation level, the very small thermal mass and the major influence of the ground temperature. Also, the interaction between the inner and outer skin has to be considered. The model we develop deals with these specificities by dividing the shelter in seven thermal zones. Both a resistive model and an airflow model connect these zones (Fig. 7). This way, the gap between the inner and outer tents is modelled in a similar way as a double skin fac¸ade, and the model may be compared to the data collected during the field measurement. All adjacent surfaces between zones, or between zones and the outside, are considered as
174
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Fig. 7. Air model.
fully glazed. The glazing photometric properties are defined based in Table 1. This way, solar transmission through the fabric is considered. Only the ground sheet is modelled as an opaque material. The thermal conductivity of the glazing is also adapted to match values specific to the various fabrics used. The airtightness of the tent is the main difficulty. Indeed the fabrics are not airtight. Further, windows and doors are closed by Velcro bands only. The connection to the ground also shows air leakages. Fig. 7 details the air model we created to deal with such behaviour. Airflows through the outer tent are related to both the temperature difference and the wind pressure. The
effect of the temperature difference is easily modelled by the Archimedes equation. The wind pressure is modelled by using pressure coefficients which are determined thanks to TNO Cp Generator [27] (see Appendix). Note that the main obstacles around the tent in the site of measurements were defined to obtain the most accurate values of wind pressure coefficient on the tent. The values we evaluated for the test case in Limelette are presented in Appendix. All openings are described by an area, a flow exponent and a flow coefficient. An indicative area for all openings is determined based on the actual geometry of the tent. To consider the air infiltration throw cracks and fabric, all glazed surface are considered as fully
Fig. 8. 4–10 October calibration: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
opened and the discharge coefficient of each glazed surface is calculated to correspond to the air permeability of the corresponding fabric and crack. Flow exponents are related to the king of opening (crack or large opening) and indicative values are easily found in the thermal software documentation. In this study, only large openings are modelled, with a flow exponent of 0.65, to represent air infiltration as well throw fabrics as throw crack. This way, the only uncertainty is about the discharge coefficient of the openings. 3. Model calibration The thermal model of the emergency family shelter is calibrated using the field measurements realised in the BBRI facility in Brussels, Belgium. The parameters to calibrate are the discharge coefficients used to model the air infiltration throw cracks and fabric of the tent and the soil thickness interacting in thermal behaviour of the shelter. To determine these coefficients, we proceed this way: firstly we simplify the model by considering that only five different values of discharge coefficient are to be identified, i.e. two for the outer tent openings (windows and vents separately), one for the mud flaps (mainly to represent crack infiltration), one for the inner tent fabric and one for the inner vents – considering only air permeability of the fabric for other surfaces (outer roof and walls), based on laboratory measurements. We evaluated also some soil thickness under the ground sheet which may be consider as part of the composition of the floor of the tent,
175
distinguishing lateral zones, transition zones and inner zone. Secondly, we make different simulations for various values of flow coefficients and soil thickness and calculate the mean variation between the simulation results and monitored temperatures for the monitored zones and especially the zone describing the inner tent. This exercise is done for the three monitoring periods in Brussels (i.e. with all openings closed and with external and internal vents opened). Some results for the first period of measurements (i.e. with all openings closed) are shown in Fig. 8. Based on all results, the difference between the calculated and the measured indoor temperatures is characterised by two polynomial equations of the third degree calculated by the statistical program JMP [28]; i.e. one for all openings closed and one for the tent with external and internal vents opened. The values of discharge coefficients and soil thickness leading to the minimum indoor temperature differences between measurements and calculations are identified by resolving the polynomial equations in excel, giving limit values of each parameter. Results are presented in Table 2. 4. Results and discussion 4.1. Quality of the model calibration Fig. 8 presents the results of calibrated tent model for the first periods of measurements in Belgium, with all opening closed. Fig. 8a presents the measured and calculated indoor temperatures
Fig. 9. 10–24 October calibration: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
176
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Fig. 10. 7–27 November calibration: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
and the outdoor temperature. Fig. 8b illustrates the hourly indoor temperature difference between calculation and measurement (sticks) and the confidence interval of this temperature difference on the considered investigation period so that the temperature difference ranges in this interval during 95% of the time. The correspondent values of discharge coefficient and soil thickness are presented in Table 2. These lead to a mean deviation between measured and calculated temperature of 1.94 ◦ C. We observe that the value of discharge coefficient of outer vents have a small impact on goodness of the model. Considering only daily period, the best ground temperature to consider is 2 ◦ C higher than for the whole period. In the model, we can only choose a constant ground temperature or a ground temperature equal to outdoor temperature. The ground temperature is thus difficult to model but
Table 2 Calibration results. Parameter
Calibrated value
Discharge coefficients Inner tent Mud flaps External windows, closed External vents, closed Internal vents, closed External vents, opened Internal vents, opened
0.0242 0.0052 0.0840 0.1595 0.1595 0.1595 0.2300
Soil thickness participating to floor composition of Transitional and inner zones Lateral zones
15 cm 5 cm
Table 3 Simulation results for calibration field measurements periods. Period of measurement
Measured temperature
[◦ C]
Calculated temperature
[◦ C]
Difference between measurement and calculation
4–10 October
Maximum Minimum Variation
31.8 6.3 25.5
Maximum Minimum Variation
28.8 2.3 26.5
3.0 ◦ C 4.0 ◦ C 1.0 ◦ C
10–24 October
Maximum Minimum Variation
29.8 2.7 27.1
Maximum Minimum Variation
28.8 2.3 26.5
1.0 ◦ C 0.4 ◦ C 0.6 ◦ C
7–27 November
Maximum Minimum Variation
23.2 1.0 22.2
Maximum Minimum Variation
28.0 1.5 26.5
4.8 ◦ C 0.5 ◦ C 4.3 ◦ C
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182 Table 4 Ground temperatures considered for simulations in Burkina Faso. Constant value of ground temperature at
[◦ C]
Building surface Deep Shallow
30 28 29
has a large impact on result, i.e. in order of +0.12 ◦ C in mean deviation per degree in ground temperature. We may also note that it is difficult to reproduce the night overcooling in case of clear sky. During the periods of 10–24 October and 7–11 November, the inner and the outer vents are opened and their discharge coefficients are presented in Table 2. The mean deviations between
177
measured and calculated internal temperatures are respectively 1.59 ◦ C and 1.77 ◦ C. Results are presented in Figs. 9 and 10. Finally, a summary of the simulation results is presented in Table 3. The model represents between 98% and 119% of the internal temperature variation during the three periods of calibration considered. 4.2. Validation of model calibration in summer conditions in Burkina Faso We may then check the calibration of the thermal model with measurements realised in Burkina Faso for the Family Tent with all openings closed. Simulation results with the calibrated model
Fig. 11. Field measurement in Burkina Faso [30].
Fig. 12. Some simulation results in Burkina Faso [30]: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
178
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Table 5 Minimum and maximum internal temperatures: measured versus calculated. Measured temperature
[◦ C]
Calculated temperature
[◦ C]
Difference between measurement and calculation
Maximum Minimum Variation
51.4 18.9 32.5
Maximum Minimum Variation
49.0 20.5 28.5
2.4 ◦ C 1.6 ◦ C 4.0 ◦ C
Table 6 Minimum and maximum internal temperatures: measured versus calculated. Measured temperature
[◦ C]
Calculated temperature
[◦ C]
Difference between measurement and calculation
Maximum Minimum Variation
28.8 −3.7 32.5
Maximum Minimum Variation
31.1 −1.1 32.2
2.3 ◦ C 2.6 ◦ C 0.3 ◦ C
are presented in Fig. 12. The in situ measurements concern the outdoor temperature and the indoor temperature of the Family Tent. No information about wind speed, ground temperature and solar radiation during the field measurement is available. Nevertheless, the weather was sunny during this period. It is thus possible to reconstruct the solar radiation based on the location of the considered shelter. The weather data considered for the simulation are based on the climate file for Ouagadougou available on EnergyPlus website [29], except for the dry bulb outdoor temperature
which is modified to correspond to field measurements. The ground temperature is defined as presented in Table 4. The construction method of the meteorological input file considered for the calculation of the internal temperature of the shelter explains mainly the differences observed between measurement and calculation. Indeed, the difference between calculated and measured internal temperature in the Family Tent (Fig. 12) is the most important from 11 to 13 November 2013. During these days, the wind speed in the climate file considered is very high. Maybe
Fig. 13. Simulation results in Luxembourg [31]: (a) measured and calculated temperatures and (b) temperatures difference between calculation and measurements.
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
179
Fig. 14. Simulation result in Brussels.
the in situ wind speed is lower than the considered wind speed for the calculation. The mean internal temperature difference between calculation and measurement for the Family Tent is 2.46 ◦ C. On the other hand, the maximum and minimum measured internal temperatures during the field measurements are respectively 51.4 ◦ C and 18.9 ◦ C and it appears respectively on 13 and 25 November. The calculated maximum and minimum internal temperatures during the field measurements are respectively 49.0 ◦ C and 20.5 ◦ C and it appears on 18 November and 2 December. The model may thus represent 87.7% of the internal temperature variation, as presented in Table 5 (Fig. 11).
The weather was not fully sunny then the solar radiation must be reconstruct on base of time of insulation and daily mean value of solar radiation for 30 years. The weather data are then not accurate and the simulation results are consequently not very good. Nevertheless, it appears that the most difficult thing to reproduce is the night overcooling in case of clear sky, as presented in Fig. 13. However, the model reproduces 99% of the internal temperature variation over the measurement period, which is pretty good, as presented in Table 6.
4.4. Solar radiation and wind speed 4.3. Validation of model calibration in winter conditions in Luxembourg [31] The field measurement in Luxembourg is difficult to reproduce due to the lack of meteorological information. Indeed, the outdoor temperature as well as the wind’s speed and direction available correspond to the airport localised at approximately 13 km at North-East of the experimentation place. Also, no solar radiation data is available excepted 30 years daily mean value of radiation and number of minutes of insulation per hour at the airport of Luxembourg.
Fig. 14 presents the complete meteorological data for the first period of measurement in Brussels, Belgium. It appears a large dependence of the indoor temperature to as well solar radiation then wind speed. The thermal model reproduced this dependence with a pretty good reproduction of thermal behaviour of the shelter during daily periods. However, the difficulty to represent night overcooling in the case of clear sky may be seen during nights from 6 to 7 October and from 9 to 10 October.
Table 7 Points of wind pressure coefficients calculation.
Fig. 15. Site of measurements in Belgium.
Calculation point
X [cm]
Y [cm]
Z [cm]
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
0 65 235 595 660 595 235 65 65 65 235 595 595 595 235 65
200 65 0 65 200 335 400 335 200 100 100 100 200 300 300 300
95 105 100 105 95 105 100 105 135 135 135 135 135 135 135 135
180
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Fig. 16. Projections for Cp calculation in Belgium.
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
181
Table 8 Cp values at each point of calculation for different wind directions. Wind direction Calculation point
0 North
45
90 East
135
180 South
225
270 West
315
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
−0.688 −0.258 −0.140 −0.258 −0.697 0.048 0.280 0.044 0.220 −0.280 −0.130 −0.030 0.244 0.036 0.214 0.034
0.037 −0.698 −0.234 −0.138 −0.258 −0.586 0.061 0.271 0.103 −0.940 −0.188 −0.740 0.103 −0.448 0.089 0.200
0.262 0.054 −0.572 −0.239 −0.133 −0.239 −0.572 0.054 −0.738 0.070 −0.391 −0.270 −0.738 −0.204 −0.859 0.080
0.037 0.271 0.061 −0.586 −0.258 −0.138 −0.234 −0.698 −0.257 −0.700 0.039 −0.390 −0.265 −0.130 −0.267 −0.940
−0.688 0.044 0.280 0.044 −0.688 −0.258 −0.140 −0.258 −0.127 −0.250 0.214 0.070 −0.133 −0.280 −0.130 −0.280
−0.248 −0.569 0.056 0.259 0.039 −0.698 −0.228 −0.138 −0.154 −0.400 0.082 0.230 −0.167 −0.939 −0.188 −0.300
−0.133 −0.239 −0.572 0.054 0.262 0.054 −0.572 −0.239 −0.301 −0.200 −0.859 0.080 −0.331 0.075 −0.391 −0.270
−0.261 −0.153 −0.233 −0.706 0.042 0.166 0.067 −0.586 0.045 −0.140 −0.268 0.240 0.051 0.170 0.039 −0.450
However, some phenomena are difficult to reproduce, as overcooling during night in clear sky conditions. Also, the ground temperature has a wide impact on results but is difficult to model. To conclude, using windows to model fabric of tent permits, by adapting discharge coefficients and photometric properties, to reproduce the properties of air permeability and luminous transmittance of fabrics and obtain a good representation of light structure thermal behaviour. Other similar models may be constructed in view to compare form, material, and design to assess the quality of such installation. Therefore, the methodology described in Section 2.3 is a methodological recommendation for further studies.
Acknowledgement Fig. 17. Positions of wind pressure coefficient calculation.
5. Conclusion The aim of this paper is to discuss the creation of a model representing the thermal behaviour of the family shelter for emergency situations used by the UNHCR, the IFRC and ICRC [24]. The main hypothesis (described in Section 2.3) considered for the model construction are: (1) the model is divided in several zones in order to represent the thermal behaviour of the gap between inner and outer skin; (2) these zones are connected with both a resistive and an airflow model; (3) all fabric surfaces (except ground sheet) are considered as fully glazed to represent the translucence of the fabrics; (4) all these glazed surfaces are considered as opened with an adapted discharge coefficient to represent the airtightness of the shelter. After calibration with field experiment data for Belgium, the model is relevant and may reproduce the actual behaviour of a tent, as well for Belgium than for other climatic conditions (as shown for Burkina Faso and Luxembourg). The deviation calculated between the measured and calculated internal temperature in Brussels, for different conditions (i.e. vents opened or closed), ranges between 1.6 and 1.9 ◦ C for the three periods of measurement considered. When the calibrated model is transposed in Burkina Faso, the value of the mean deviation between measured and calculated indoor temperature is 2.5 ◦ C and the model reproduces 87.7% of the internal temperature variation (Tmax –Tmin ) within the considered period. Respectively, when considering winter condition as evaluated in Luxembourg, the mean deviation over the field measurement period is around 2.5 ◦ C and the model reproduces 99% of the internal temperature variation.
This research was partly conducted in the frame of the SIMBA FEDER project, supported by the Walloon Region and the European Union.
Appendix A. Appendix A.1. Cp values in Limelette The values of wind pressure coefficients are determined thanks to TNO Cp Generator [27]. The values corresponding to the test case in Limelette are presented in Table 8. Note that the main obstacles around the tent in the site of measurements were defined to obtain the most accurate values of wind pressure coefficient on the tent. The assumptions taken for the evaluation of Cp values with the Cp Generator are as follows: - The environment is defined as cultivated land, that is to say a plain with small obstacles (the obstacle distance is greater than twenty times the height of the obstacle): z0 = 0.5 m; - The weather station of the BBRI was taken as reference (about 30 m from the shelter) with a height of the wind speed measurement of 12 m; - The grove, the tree row and the house are modelled as a major obstacles (Fig. 15). With these assumptions, the height considered to calculate the Cp values is the height of wind speed measurement (12 m) and the local roughness at the measuring point is regarded as identical to the surrounding terrain. This allows to calculate the wind pressure directly from local wind data without correction. In other words, the Cp values are corrected for the data of the local weather station.
182
S. Obyn et al. / Energy and Buildings 89 (2015) 170–182
Note also that only rectangular building may be introduced in the TNO Cp Generator. The trick is then to make rectangular projections of the tent that approach the blockage of the wind. There are three typical facades: - Long side; - Short side; - Oblique side. The other five sides may be derived from cyclic substitution. Hence, three different projections are realised, as shown in Fig. 16. The results of Cp value on each fac¸ade (Fig. 17) are given in Tables 7 and 8. References [1] J. Ashmore, Tents: A Guide to the Use and Logistics of Family Tents in Humanitarian Relief, United Nations Publication, 2004, OCHA Ref Nr. OCHA/ESB/2004/19. http://josephashmore.org/publications/tents.pdf. [2] UNHCR, Handbook for Emergencies, United Nations High Commissioner for Refugees, Geneva, 2000, pp. 405. [3] P. Foley, UNHCR – Shelter Programme Monitoring and Evaluation: Final Report, 2005, pp. 6. [4] IFRC – SRU Field Tests & Findings Research and Development Project Launched by IFRC IRC UNHCR on Tents Investigations, Lux Red Cross, Bujumbura, Burundi, July–Augustus 2013. [5] L. Alegria Mira, A.P. Thrall, N. De Temmerman, Deployable scissor arch for transitional shelters, Autom. Constr. 43 (2014) 123–131. [6] P.J. Owens, A. Forgione Jr., S. Briggs, Challenges of international disaster relief: use of a Deployable Rapid Assembly Shelter and Surgical Hospital, Disaster Manag. Response 3 (1) (2005) 11–16. [7] P. Manfield, A comparative study of temporary shelters used in cold climates. What can be learnt from the design of the Yurt and the Scott tent to inform the future design of shelters systems for emergency relief? in: History Essay for the MPhil Degree in Environmental Design in Architecture, Cambridge University, Cambridge, UK, 2000 http://www.shelterproject.org/ downloads/manfield1.pdf [8] P. Manfield, Modelling of a Cold Climate Emergency Shelter Prototype and a Comparison with the United Nations Winter Tent. Technical Essay for the MPhil Degree in Environmental Design in Architecture, Cambridge University, Cambridge, UK, 2000 http://shelterproject.org/downloads/manfield2.pdf [9] M. Asefi, F. Ahangar Sirus, Transformable shelter: evaluation and new architectural design proposals Procedia Soc. Behav. Sci. 51 (2012) 961–966. [10] J. Roekens, V. Virgo, L. De Leat, M. Van Craenenbroeck, S. Puystiens, M. Mollaert, A new type of sheltering for disaster affected sites: the ‘Clever roof’, in: Proceedings of the International Association for Shell and Spatial Structures (IASS) Symposium 2013, Wroclaw University of Technology, Poland, 2013. [11] C. Henrotay, A contribution to an integrated and more sustainable design approach for the material support of shelter after a disaster (Ph.D. dissertation), Vrije Universiteit Brussel, Brussels, 2008. [12] R. Battilana, Design of Cold Climate Temporary Shelter for Refugees, 2000 http://www.shelterproject.org/downloads/cold%20climate%20liner.pdf
[13] C. Hale, R. Dooley, Keeping People Warm: Collaborative R&D, ShelterBox, 2014 http://www.shelterbox.org/uploads/Sources/Downloads/shelterbox trust annual report 2012.pdf [14] C. Henrotay, M. Mollaert, H. Hendrickx, Adaptable Construction Systems for Shelter in Emergency Situation. Adaptables 2006 Proceedings of the Joint CIB, Tensinet, IASS International Conference on Adaptability in Design and Construction, vol. 2, No. 4, 2006, pp. 63–67. [15] C. Henrotay, W. Debacker, M. Mollaert, W. De Wilde, H. Hendrickx, 4Dimensional Design: A Design Strategy for Efficient Shelter and Sustainable Housing After Conflict-based and Natural Disasters. Management of Natural Resources, Sustainable Development and Ecological Hazards, vol. 1, WIT Press, 2006, pp. 341–350. [16] C. Crawford, P. Manfield, A. McRobie, Assessing the thermal performance of an emergency shelter system, Energy Build. 37 (5) (2005) 471–483. [17] X.W. Winston, R.E. England, K.C. Kuivinen, J.J. Potter, N.S. Krug, A critical review of design and use of field tent shelter in polar regions, Polar Rec. 34 (189) (1998) 113–122. [18] P. Manfield, T. Corsellis, Cold Climate Emergency Shelter Systems. A Research Project for Humanitarian Organisations, Unpublished Paper for the Halley Stewart Trust, Cambridge, UK, 1999 http://www.shelterproject.org/ downloads/coldshelter.pdf [19] P. Manfield, Emergency Shelter for Humanitarian Relief in Cold Climates: Policy and Praxis, Unpublished Dissertation for a Diploma in Architecture, Martin Centre for Architectural and Urban Studies, Department of Architecture, Cambridge University, Cambridge, UK, 2001 http://www.shelterproject.org/ downloads/coldshelter2.pdf [20] J. Rudge, R. Gilchrist, Measuring the health impact of temperatures in dwellings: investigating excess winter morbidity and cold homes in the London Borough of Newham, Energy Build. 39 (7) (2007) 847–858. [21] H. Johnson, R.S. Kovats, G. McGregor, J. Stedman, M. Gibbs, H. Walton, The impact of the 2003 heat wave on daily mortality in England and Wales and the use of rapid weekly mortality estimates, Eurosurveillance 10 (7–9) (2005) 68–171. [22] J. Krieger, D.L. Higgins, Housing and health time again for public health action. Public health matters, Am. J. Public Health 92 (May (5)) (2002). [23] J. Damerell, T. van Zutphen, The Sphere Project: Humanitarian Charter and Minimum Standards. Humanitarian Response, Edition, 2011, ISBN: 978-1908176-004, http://www.ifrc.org/docs/idrl/I1027EN.pdf [24] UNHCR (United Nations High Commissioner for Refugees), Family Tent: UNHCR Item No. 05353, 2011. [25] DesignBuilder (Version 3.2.0.073) © Copyright 200-2013, DesignBuilder Software Ltd. [26] D.B. Crawley, J.W. Hand, M. Kummert, B.T. Griffith, Contrasting the capabilities of building energy performance simulation programs, Build. Environ. 43 (2008) 661–673. [27] Nederlandse Organisatie voor Toegepast Natuurwetenschappelijk Onderzoek (TNO). TNO Webapplications, http://cpgen.tno.nl/cp/info.asp [28] JMP® 10.0.0 Edition en 64 bits, Copyright © 2012 SAS Institute Inc. All Rights Reserved. [29] U.S. Department of Energy Website. http://apps1.eere.energy.gov/buildings/ energyplus/weatherdata request search.cfm?sortKey=country&opt=1, accesses 01.08.14. [30] IFRC–SRU Field Tests Investigations Tough Speedkits Project, Comparison Between Standard Family Tent With and Without Shade Net, Sag Nioniogo, Burkina Faso, November 2013. [31] IFRC–SRU Field Tests & Findings, Research on Winterization Potential Solutions, IFRC–SRU Field Tests & Findings, Bertrange, Luxembourg, March 2014.