THERMAL COMFORT SIMULATION IN MODERN AIRCRAFT COCKPITS Paul Mathis 1, Rita Streblow 1, Dirk Müller 1, Gunnar Grün², Andreas Wick³, Jean-Christophe Thalabart³ 1 Institute for Energy Efficient Buildings and Indoor Climate, E.ON Energy Research Center, RWTH Aachen University, Germany. 2 Fraunhofer Institute for Building Physics, Valley, Germany. 3 Airbus Operations GmbH, Hamburg, Germany. Abstract The non-uniform thermal environment in an aircraft cockpit during standard cruise conditions is simulated for a specified ventilation configuration and three different boundary condition sets. The air change efficiency is increased by changing the wall boundary condition from adiabatic to a specified temperature value and additionally, by installing air outlets in the ceiling. Due to strongly inhomogeneous flow conditions, locally resolved assessment of thermal comfort is required. The thermoregulatory answer of the pilots and a 3 rd occupant is simulated by a multi-compartment thermal comfort model providing local and overall thermal sensation and comfort values. Local cold discomfort at outward pointing shoulders and feet is identified for the pilots resulting in moderate overall thermal comfort. The 3 rd occupant is not expected to suffer from poor thermal comfort. Keywords: aircraft cockpit, CFD, thermal sensation, thermal comfort, thermal comfort model 1 Introduction Thermal comfort is an issue of growing importance in modern aircraft development. The cockpit is a zone of special interest, because poor thermal comfort is expected to lead to decreased alertness and productivity as shown by Wyon et al. (1996). Thus, improvement of the thermal environment may help minimizing risks in aviation due to human failure. In the last years, emphasis on thermal comfort has widened from the passenger cabin only to more operational regions like the cockpit, galleys and lavatories. However, investigations published concentrate on thermal stress and comfort issues in military aircraft cockpits (Jain et al., 2001). In this study, thermal comfort is investigated numerically in a modern commercial aircraft cockpit. Extreme thermal boundary conditions have to be considered when dealing with thermal comfort problems in aircraft cockpits, e.g. high thermal loads due to instruments, occupants and solar radiation, heat transport between interior and exterior, where temperatures can range from +50 C on ground down to -50 C and even lower during cruise. Moreover, due to smoke removal requirements in emergency cases, the cockpit is supplied with a high volume flow rate. In a first step, the thermal environment is simulated with Computational Fluid Dynamics (CFD). Cockpit airflow is naturally time-dependent, turbulent and partially buoyancy-driven. All these effects, as well as radiation, are modeled. The commercial simulation environment ANSYS Workbench including the CFX flow solver is used for the simulation chain. In a second step, the thermoregulatory response of the human body is simulated with a multi-compartment thermal comfort model, allowing to assess the occupants thermal sensation and comfort locally and globally.
2 Methods 2.1 Simulation of the Thermal Environment with CFD The computational domain consists of the cockpit volume, where two pilots and a 3 rd occupant are seated (Fig. 1). To the bottom, it is limited by the avionics bay compartment and to the by a wall separating it from the passenger area. Due to complex geometry, a mainly unstructured mesh is used (Fig. 1). The major part of the flow domain containing the curved geometry is discretized with tetrahedrons (7.95 million cells), whereas core regions are discretized with hexahedrons (0.19 million cells) and connected to the tetrahedron region through pyramids. At all heat Figure 1: Geometry and mesh transferring boundaries (cockpit walls, instrument panels, occupants) the boundary layer is resolved with 5 prismatic layers (1.85 million cells), so that y + 1 is ensured. This amounts to a total of about 10 million cells (2.57 million nodes). In all calculations, a typical standard cruise situation is considered (table 1). The cabin pressure amounts to p ref = 8 10 4 Pa leading to a reference density of 0.94 kg/m 3. Because of mixed convection, buoyancy is taken into account. The time-dependent behavior is captured by the transient 2 nd order ward Euler scheme with a time step of t = 0.5 s. The time step was determined from a sensitivity study. After reaching thermal equilibrium, a total time of 180 s is calculated for averaging purposes. Turbulence is modeled with the URANS k-ω-bsl model developed by Menter (1994) and thermal radiation is modeled with the discrete transfer model. These models were found to be accurate for mixed convection due to heat sources by Streblow (2011). For the estimation of the age of the cockpit air, an additional transport equation for the variable age of air is solved. Fresh air is supplied to the cockpit through symmetrically arranged air inlets fulfilling the requirement of 283 l/min per occupant for the maximum of 4 occupants in the cockpit. The temperature of the supply air amounts to 24 C. Each of the three occupants delivers a thermal load of 100 W. The instrument panels in the front part, between the pilots and above the pilots s deliver a total heat of 45 W. All wall boundaries (cockpit walls, windows, seats, occupants) are modeled as opaque surfaces with an emissivity of 1, except the panels with an estimated emissivity of 0.9. The particular temperatures at the wall boundaries are given in table 1. Three different configurations with regard to the boundary conditions are presented in this work (table 1), whereas the ventilation configuration with regard to supply volume flows is constant. Case A: Adiabatic walls. In this case, all cockpit walls are expected to be adiabatic. That would imply perfect insulation of the aircraft against external temperatures typically around -50 C during cruise. Thus, it is a theoretical case meant to show the impact of perfect thermal insulation. Case B: Specified temperature on walls. In this case, wall temperatures are derived from flight test data. Thermographic analyses indicate a wall temperature of 23 C throughout the cockpit lining and lower temperatures at window frames.
Case C: Specified temperature on walls with outlets at ceiling. During the progress of research, air outlets in the ceiling region appeared reasonable. Thus, in case C additional ceiling air outlets in the rear upper part of the cockpit are integrated. Table 1: Models and boundary conditions Models Boundary Conditions Fluid air ideal gas Occupants 100 W each Analysis type Transient Instrument panels 45 W total Time step 0.5 s Air inlets > 1132 l/min total Averaging time 180 s Windows 35 C Reference pressure 80000 Pa Window frames Case A: adiabatic Reference density 0.9385 kg/m³ Case B/C: < 23 C Radiation discrete transfer Walls Case A: adiabatic Turbulence k-ω BSL Case B/C: 23 C 2.2 Simulation of Thermal Comfort Thermal comfort is modeled with a Modelica-based code developed by Streblow (2011). The thermal comfort model is a multi-compartment model dividing the human body into 16 parts and thus allowing local assessments (Fig. 2). From the thermal environment, the interior processes and thermoregulatory responses of an actual human body are simulated in a physiological model. Thermal sensation and comfort are predicted from the variation of physiological conditions throughout the body by a psychological model. The model has been calibrated by experimental data. A coupling between comfort modeling and CFD is possible in order to capture the influence of the calculated thermal conditions of the human body on the Figure 2: Segmented body flow (dotted arrow in Fig. 3). The work flow is illustrated in Fig. 3. The non-uniform thermal environment is simulated with CFD methods. Average heat transfer coefficients due to convection and radiation are extracted for all body parts. Also, the average environmental temperature and the average radiation temperature are derived from CFD calculations. Moreover, the relative humidity, the clothing value for each body part and the metabolic rate are supplied as boundary conditions for the physiological model. Figure 3: Work flow of thermal comfort simulation
In the physiological comfort model, the human body state is calculated based on the human heat balance equations. Each body part consists of a core and a skin layer. Also, a central blood compartment is integrated, making a total of 33 nodes. For each node, the heat balance is computed. All major heat transport processes like heat production through external work, basal metabolism, shivering, interior heat conduction, blood flow, heat loss through respiration and evaporation, convection and radiation at skin surface, conduction at contacted surface and vasomotion are simulated. In a psychological model, the physiological body state in terms of local temperatures is transformed into local thermal sensation. From local thermal sensation, an overall thermal sensation value is derived by weight functions specific for each body part. Local thermal comfort depends on local and on overall thermal sensation. It is derived from an asymmetrical saddle function that takes into account both local and overall thermal sensation. Finally, an overall thermal comfort index is derived using weight functions from local thermal comfort values. All sensation values are indicated on the commonly used 7-point ASHRAE scale (ASHRAE Standard 55, 2009) ranging from -3 (cold) over 0 (neutral) to +3 (hot) and the comfort values are indicated on a 6-point scale ranging from -3 (very unpleasant) to +3 (very pleasant). In this work, the emphasis is laid on the modeling of the thermal environment with CFD. The thermal comfort model is used for post-processing the data and thus assessing the thermal comfort. 3 Computational Results 3.1 CFD Results Prior to computing results, a mesh study has been conducted to ensure mesh independency. The mesh refinement on the surface of the occupants as well as the prism layer growth rate was varied. It has been found that a medium sized grid (10 mm edge length) combined with a growth rate of 1.75 for the 5 prismatic layers producing a smooth transition of cell size to the outer mesh gives similar results to those of a fine mesh (5 mm edge length, growth rate 1.5). Thus, a suitable compromise between accuracy and computation time has been found. The air change efficiency is normalized by the value calculated for case A. For case B, the air change efficiency is 1.2 times larger compared to case A and 1.7 times larger for case C, respectively. In Fig. 4 and 5, temperature and velocity are plotted on cross sections y = 0.5 m (pilot) and y = 0 m (3 rd occupant). In Fig. 6, the age of air is given on cross section y = 0.5 m. All values are averaged over 180 s simulation time. In Fig. 4, the results of the averaged temperature fields are shown. For case A, a temperature gradient of about 6 K between foot and area is observed. For cases B and C, the temperature gradient amounts to about 2 K. Above the 3 rd occupant, thermal plumes are visible. In cases B and C, the temperatures in the wall region are about 24 C due to the influence of the wall temperature. Close to the windshields the temperatures are above 30 C. In Fig. 5, the results of the averaged velocity fields are shown. In general, the level of velocity magnitude is lowest for case A, higher for case B and highest for case C. In large parts of the domain, the flow is chaotic. In case B and C, a recirculating flow pattern along the cockpit ceiling and wall is observed. Moreover, larger velocities pointing to the rear upper part of the cockpit at y = 0 m are visible for cases B and C compared to case A. Also, thermal plumes corresponding to Fig. 4 are observed. In proximity of the pilot, velocities greater than 0.5 m/s occur. In Fig. 6, the results of the averaged age of air are shown. Generally, in the front part of the cockpit the age of air is lower than in the upper rear part for all cases. From case A to case C the highest value for age of air on the shown cross section is reduced from about 180 s in case A to about 120 s in case B and about 70 s in case C. Near the front windshields, the air is also older compared to the other parts of the cockpit. In the region of the pilot s, the age of air is 40 to 50 s for cases A and B and 30 to 40 s for case C.
Case A Case B Case C Figure 4: Temperature plots on cross sections y = 0.5 m (top row) and y = 0 m (bottom row). Case A Case B Case C Figure 5: Velocity plots on cross sections y = 0.5 m (top row) and y = 0 m (bottom row). Case A Case B Case C Figure 6: Age of air plots on cross section y = 0.5 m.
3.2 Thermal Comfort Modeling Results In this section, predicted thermal sensation (PTS) and predicted thermal comfort (PTC) results are shown. In the graphs in Fig. 7, the pilot sitting on the left side is referred to as CA (captain) and the one sitting on the right side as F/O (flight officer), respectively. In all cases, the captain s and flight officer s PTS for most body parts lies within a range of ±0.5 indicating a neutral state. Exceptions are found with respect to the shoulder and arm: The captain s left shoulder and the flight officer s right shoulder indicate a PTS of about -1 in case A and of about -1.6 in cases B and C. Their overall PTS is nearly neutral for case A and about -0.4 for cases B and C. The 3 rd occupant s local PTS values lie within a range of ±0.5 in all cases except for the upper body parts ( down to the arms) in case A, where values between 0.5 and 1.3 are found. His overall PTS is about 0.9 for case A and about -0.2 for cases B and C. 2.0 1.5 1.0 0.5 0.0-0.5-1.0-1.5-2.0 hot cold Case A: Thermal Sensation CA F/O 3rd Case B: Thermal Sensation Case C: Thermal Sensation 2.0 1.5 1.0 0.5 0.0-0.5-1.0-1.5-2.0 pleasant unpleasant Case A: Thermal Comfort CA F/O 3rd Case B: Thermal Comfort Case C: Thermal Comfort Figure 7: Local and overall thermal sensation (upper row) and comfort (lower row). With regard to thermal comfort, greater overall PTC for the pilots is estimated for case A (0.6) compared to cases B and C (0.2). Their most comfortable body parts are the, the and the hands in all cases and in cases B and C the arms pointing inwards to the cockpit (CA: right arm, F/O: left arm). The lowest PTC values are found for the shoulders pointing outwards (CA: left shoulder, F/O: right shoulder): about 0.2 in case A and about -0.5 in cases B and C. Moreover, in cases B and C the PTC values for the feet are between -0.3 and 0. The overall PTC for the 3 rd occupant is about 0.7 for all cases. Like for the pilots, the body parts with highest PTC are the, the (except for case A) and the hands. The lowest PTC values are found for the arms in case A and the feet in cases B and C. 4 Discussion 4.1 Comparison of flow structures The air change efficiency rate is dependent on the air flow pattern. Most of the fresh air is supplied in the front part of the cockpit and the air is extracted in the floor region. In case A, due to adiabatic walls and buoyancy, the heated air is accumulated in the rear upper region. This is also
indicated by a higher level of age of air (Fig. 6) and a lower velocity level than in the other cases (Fig. 5). In case B, a temperature smaller than the resulting air temperature is specified on the walls and window frames (table 1). That causes the air adjacent to the walls to cool and follow the gravitational force due to buoyancy. By that, a recirculating air pattern is observed. Due to stronger mixing compared to case A, the air change efficiency is increased by 20 %, the vertical temperature gradient is decreased from 6 K to 2 K and the age of air is decreased as well (Fig. 4 to 6). Besides the specified wall temperatures in case B, additional air outlets in the rear part of the ceiling are integrated in case C allowing the relatively old air being extracted more easily. Thus, the velocity level and mixing are increased, the vertical temperature gradient is decreased even more and also the age of air is decreased (Fig. 4 to 6), totaling in an air change efficiency increase of 70 % compared to case A. 4.2 Comparison of thermal comfort In all cases, the predicted thermal sensation (PTS) and predicted thermal comfort (PTC) values show a symmetrical structure with regard to body parts such as shoulders and arms due to the symmetric pattern of wall heat transfer coefficients (Fig. 8). This pattern is caused by symmetric inlet flows and mainly symmetric geometry and thermal boundary conditions. Case A Case B Case C Figure 8: Total wall heat transfer coefficient at the occupants surfaces. Considering case A, although comfort values according to ASHRAE 55 are violated by e.g. velocity magnitudes larger than 0.5 m/s and a vertical temperature gradient of 6 K, the overall PTC for all three occupants lies in an acceptable range of higher than 0.6. According to the comfort model, the resulting high wall heat transfer at the,, shoulders and arms is tolerated by the pilots. An explanation for that may be the tendency of humans to accept higher velocities in an environment with high ambient temperature and low humidity (in this case 30 C and 15 % relative humidity), as found by Arens et al. (2009). Unlike the pilots, the 3 rd occupant is not exposed to any air jet coming from an inlet directly. For his arms and feet lowest PTC values are calculated due to high ambient velocities impacting the overall PTC, so that a value of about 0.7 is calculated. With regard to thermal sensation and comfort, cases B and C are similar. This results from the similar pattern of the thermal and flow variables in close proximity of the occupants (Fig. 4 to 6 and 8). The low PTS values, especially those of the pilots outward pointing shoulders and feet, contribute to the poor overall PTC of 0.2. The shoulders pointing outwards and the feet are hit by supply air jets resulting in high wall heat transfer coefficients (Fig. 8). Due to the lower ambient temperature, the impact of low local PTC values is stronger compared to case A. The 3 rd occupant s feet are exposed to high ambient velocities resulting in the poorest local PTC values in his case. His overall PTC results to 0.7. Generally speaking, body parts shadowed from large wall heat transfer coefficients i.e. inward oriented shoulders and arms, the and the thighs lead to acceptable PTC in all cases. Furthermore, due to the averaging process in order to obtain a single value for the wall heat transfer coefficient for a whole body part, peak values are alleviated and thus, draft risk is eventually underestimated for example for body parts like the and the (Fig. 8).
5 Conclusion and Outlook Flow simulations show a strong impact of the cockpit walls thermal boundary condition on air change efficiency and temperature stratification. Well insulated walls (adiabatic) as simulated in case A would require a repositioning of inlets and/or outlets to ensure desired ventilation efficiency. Realistic insulation as simulated in cases B and C increases the effects of natural convection within the cockpit and leads to a better mixing of the air. Air change efficiency is improved from case A to case C by 70 %. Thermal sensation and comfort results fit to the expected tendencies due to the simulated thermal environment. In the considered ventilation configuration, main draws of thermal comfort for the pilots have been identified: draft risk at outward pointing shoulders and the feet. The predicted discomfort due to draft risk at the shoulders is furthermore intensified by a cold local thermal comfort. The 3 rd occupant is not expected to suffer from poor thermal comfort. CFD results are going to be validated with detailed measurements at a full-scale test facility being presently built at the Fraunhofer Institute for Building Physics. In the test facility, also subject tests will be performed to validate the comfort simulations and to extend the empirical data base for the thermal comfort model. 6 Acknowledgement This work has been financed by the Fraunhofer Institute for Building Physics with funds from the German Bundesministerium für Wirtschaft und Technologie under support code 20K0905M. The authors are responsible for the contents of this publication. 7 References ANSI/ASHRAE Standard 55-2009, Thermal Environmental Conditions for Human Occupancy, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. ANSYS Academic Research, Release 12.1, Help System, ANSYS, Inc. Arens, E., Turner, S., Zhang, H. and Paliaga, G. (2009) A Standard for Elevated Air Speed in Neutral and Warm Environments, ASHRAE Journal, May 51 (25), 8-18 Jain, P.K., Chawla, A. and Tyagi, P. (2001) Assessment of cabin conditioning system in a fighter aircraft, Indian Journal of Medicine, 45(2), 37-46 Menter, F.R. (1994) Two-equation eddy-viscosity turbulence models for engineering applications, AIAA-Journal, 32(8), 1598-1605 Streblow, R. (2011) Thermal Sensation and Comfort Model for Inhomogeneous Indoor Environments, PhD thesis, RWTH Aachen University Streblow, R., Müller, D., Gores, I. and Bendfeldt, P. (2009) Prediction of Thermal Sensation Using an Optimized 33 Node Simulation Model, Proceedings of the 11th International ROOMVENT Conference, May 2009, 1047-1053 Wyon, D.P., Wyon, I. and Norin, F. (1996) The effects of moderate heat stress on driver vigilance in a moving vehicle, Ergonomics 39, 61-75