Issue 1/2004
01/02/04
Final report, Thermolit door panel development
Thomas E. Lernet
In the course of a research project carried out by the MEILLER Aufzugtüren GmbH, in cooperation with the Chair for Lightweight Construction at the Technical University of Munich, an innovative door panel was developed, distinguished in particular by its excellent dimensional stability when exposed to direct sunlight.
Category: Issue 1/2004
Posted by: Editor
1.The problem at hand
Malfunctions during operations are often encountered in hoistway doors where the side facing the landing is exposed to strong, direct sunlight, this paired with high outdoor temperatures. The reason is the severe temperature differential between the shell facing the landing and the one facing the hoistway. The resulting differences in the extent of thermal expansion can cause such severe curvature that the door panels bind.
The purpose of the research project was to examine and evaluate various options and corrective measures with which the problem described here might be alleviated.
2 Potential corrective measures
2.1 Passive measures
• Increase the thickness of the material used for the stiffening beams inside the door panels.
• Using materials such as carbon fiber laminates, with far lower coefficients of expansion, inside the door panel.
• Finishing the inside surfaces of the door shells with enamel paint to improve heat transmission through radiation.
• Applying additional cladding to the door, thus enhancing the insulation effect.
• Installing thermal transfer profiles in order to reduce the temperature differential between the inside and the outside shells.
• Increase the thickness of the material used for the stiffening beams inside the door panels.
• Using materials such as carbon fiber laminates, with far lower coefficients of expansion, inside the door panel.
• Finishing the inside surfaces of the door shells with enamel paint to improve heat transmission through radiation.
• Applying additional cladding to the door, thus enhancing the insulation effect.
• Installing thermal transfer profiles in order to reduce the temperature differential between the inside and the outside shells.
2.2. Active measures
• Installing heating elements inside the door panel, on the side toward the hoistway.
• Inducing forced convection with fans in the door panel.
• Inducing forced convection with fans in the door panel.
3 Finite element calculations as the basis for decisions
3.1 Description of the model
To calculate the door panel’s response to heat, a parameterized model was created using the ANSYS finite element program (see Figures 1 and 2). Any of the dimensions, including locations, thicknesses and materials properties, can easily be modified. The model was created to represent a double-walled, sheet metal door panel 4 m tall.
To simulate the sun’s radiation for thermal calculation purposes, a surface was set up which radiates heat at output of 1 kW/m2. This value corresponds to maximum thermal impact of the sun shining perpendicular to the door surface. Consequently the calculated temperature differentials and thus the temperature-induced deformations in the model are considerably greater than what would be encountered in the real world. The actual temperature distribution due to the sun’s radiation would have to be determined in trials in order to verify the results of the calculations. A specimen door panel 1 m tall was tested (see Section 5.3)
3.2 Calculations
The results of the calculations are summarized below; these were used as the basis for identifying the most suitable solution for the problem described at the outset.
3.2.1 Basic model
The temperature distribution depicted in Figure 3 was found for the basic model. One sees that heat is transferred to the rear shell of the door panel in the area around the beams. The temperature differential is so great that, due to the differing degrees of thermal expansion, the door panel is deformed as shown in Figure 4. Maximum deformation here is 18.8 mm.
Since the temperature distribution patterns are essentially quite similar throughout, the only ones shown graphically below are those in which major deviations from the basic model can be recognized. We will forego showing the associated deformation diagram.
3.2.2 Increasing the thickness of the material used for the stiffening beams
Doubling the material thickness, to 3 mm, was examined. Due to the very limited increase in the thermal transfer rate, the temperature differential is reduced by only 2 K. Together with the increased stiffness of the beams, however, deformation is reduced to 16.3 mm.
3.2.3 Carbon fiber laminate
The influence exerted by carbon fiber laminate results from this material’s very low coefficient of thermal expansion; the intention is to prevent elongation at the outside shell when it is exposed to heat. The temperature differential rises, due to the poor thermal conductivity of the carbon fiber laminate, by 1 K. Using a laminate 3 mm thick reduces the deformation path to 8.6 mm.
3.2.4 Finishing the insides of the door shells
Further examinations were conducted to determine the extent to which heat exchange inside the door panel could be enhanced through increased thermal radiation. When using galvanized steel sheet the emission coefficient is 0.25. By comparison, this value is 0.92 for a surface finished with white enamel paint. The temperature differential is lowered by 7 K, reducing deformation to 15.8 mm.
3.2.5 Installing an additional shell on the door panel
Simulating the outside door surface when augmented with a sheet of stainless steel requires a detailed cross-sectional model. Figure 5 shows the temperature distribution for the basic configuration while Figure 6 shows the temperature distribution where there is an additional outside shell. Here the influence of the adhesive between the stiffening beams and the door shell was taken into account. The temperature differential was reduced by only 2 K.
3.2.6 Installing thermal transfer profiles
The additional models were used to examine how thermal transfer profiles might improve the transfer of thermal energy to shell facing the hoistway. Various geometries and material thicknesses were calculated here. The maximum temperature differential is lowered by 5 K; thus the deformation can be reduced to 13.5 mm.
4 Discussion of the results and selection of solutions to be further scrutinized
In the variant with stronger stiffening beams (Section 3.2.2) the reduction in temperature-induced deformation was not great enough to merit further examination, particularly since the increase in structural mass was not proportional to the attainable reduction in deformation.
A significant reduction of temperature-induced deformation can be achieved by using a carbon fiber laminate inside the door (Section 3.2.3). But the costs for this material, at about € 1,500 per door panel, are very high and thus not economically justifiable.
Good thermal transfer can be achieved where the interior surfaces are coated with enamel paint. It is, however, not sufficient for use as the sole corrective measure and would have to be combined with another alternative. Coating with enamel paint can thus be deemed a supplementary corrective measure.
Cladding the outside of the door panel with stainless steel can be disregarded due to the very limited reduction in the temperature spread.
Incorporating thermal transfer profiles (Section 3.2.6) has a beneficial effect on temperature equalization and thus reduces temperature-induced deformation. This is due, among other factors, to the fact that the temperature differential across the profiles themselves is very low, so that they curve significantly less in response to thermal expansion.
Active measures such as those suggested in Section 2.2 were discussed in depth. On the one hand, their use requires expenditures for electricity and regulation circuitry which are by no means insignificant. On the other hand, maintenance and repair of the components installed inside the door panel are rendered very difficult.
It was for these reasons that a solution involving the use of thermal transfer profiles was favored.
5 Using thermal transfer profiles
The calculation model was expanded appropriately in preparation for determining the effect of using thermal transfer profiles. Parameters for use in the model are set up for the geometries which have already been designed for economy in manufacture. The dimensions can thus be quickly optimized.
5.1 Examining the joining technique used between the thermal transfer profile and the door shell
In consideration of the differing longitudinal expansion rates for the thermal transfer profiles in comparison to the steel door shell, various joining techniques such as clamping, Tox fasteners, riveting and gluing were examined. The thermal transfer profiles were linked into the remaining FE model by way of so-called constraint equations.
The difference in the deformation when using the various joining options was of hardly any significance. Thus it was deemed to make sense to retain the joining techniques which have already become established in manufacturing.
5.2 Investigation of the number of thermal transfer profiles to be used
Further models were used to examine whether a greater number of profiles might improve the deformation properties. To ensure that the structural mass does not rise to the same degree, the shell thicknesses are selected so that the overall mass remains constant when compared with the original door panel. In order to verify the model once again, a further calculation was carried out using steel as the material for the thermal transfer profiles. The deformation was – as expected – very high.
5.3 Specimen door panel
A specimen door panel was fabricated in order to examine the calculation models and the idealizations which they contain.
5.3.1 Trial configuration
The specimen door panel has an overall height of one meter. The thermal transfer profiles are distributed uniformly across the door shell. A total of six temperature sensors and four strain gauges are installed within the cross-section, halfway up the door.
The test configuration is shown in Figure 9. The specimen door panel is hung at standard suspension points in a stiff frame. Heat is applied from a distance of about 30 cm by means of a gas-fired radiation heater, aimed at the center of the door’s front shell. This distance was selected so that a temperature of 65°C is induced at the temperature sensor at the center of the door’s front shell. Figure 10 shows the course of the temperatures at the six measurement points during the heating and cooling phases. The heating process was continued until a static situation was achieved. The cooling phase was intended to supply additional data used to evaluate the measurement system. The strain gauges were used to measure only very small amounts of expansion. No significant findings could be gained in this way.
5.3.2 Results of the trials
Observing the temperature curves, we see that temperature rises most severely at the center of the side facing the heat source. Following a certain delay the temperatures at the measurement points on the thermal transfer profiles rise and, after a further lag, the temperature at the center of the rear door shell also climbs. As was previously determined in the calculation model, the temperatures at the measurement points on the thermal transfer profiles are quite densely clustered, seen from the shell facing the heat source and continuing to the rear shell. The differences between the measurement points is explained by the asymmetry along the cross-section of the door panel. During the cooling phase the temperatures converge within a very short period of time; as expected, the temperatures are somewhat higher than at the outside. The behavior thus testifies to a reliable test configuration.
5.3.3 Comparison calculation
For the purpose of the comparison calculation a model was generated to correspond to the specimen door panel. To simulate the thermal radiator, the size of the surface otherwise used to represent the impingement of the sun was reduced. Assigned to this surface was the temperature at the thermal radiator as was measured with a temperature sensor during testing.
One beneficial aspect is that the temperature differential determined by way of calculation is greater than the value measured experimentally. This allows us to conclude that the idealization of the heat migrations between the door shell and the thermal transfer profile is correct. The deviation is due to the lack of modeling for the volume of air inside the door panel. The additional amount of heat exchange which thus becomes effective by way of thermal transfer and convection further reduces the temperature differential, and it was possible to verify this in a calculation-intensive model which was prepared once only. Here the difference between the measurement and calculation was just 2%. Consequently the deformation path predicted by calculations is greater than the value measured experimentally, as was expected.
The findings here can contribute to evaluating the calculation results discussed in Sections 5.1. and 5.2. These are also to be seen as conservative estimates, as was assumed at the outset; this means that the deformations which they predict are greater than what will be found in reality.
5.4 Optimizing the cross-sectional dimensions of the thermal transfer profiles
Optimization was subsequently undertaken using the models which had then been verified and evaluated. The goal of optimization was to minimize temperature-driven deformation while maintaining a defined maximum weight for the thermal transfer profiles inside the door panel. The design variables which were available for variation were the thicknesses of the flanges and webs and the heights and widths of the flanges. The mass was used here as a so-called “restriction,” which could not be violated. The target function was to observe and minimize maximum deformation at the center of the door panel.
The first attempt undertaken was to identify a solution using the differential-based optimizer integrated into ANSYS; this did not culminate in success for the optimization task at hand. After switching to a different optimizer, one based upon approximation models, it was possible to identify plausible optimum configurations.
6 Summary
The results in Section 5 show that temperature-related deformation declines as the number of thermal transfer profiles located inside the door panel rises. Due to cost considerations it does not make sense to install more profiles than are actually needed to ensure reliable functioning of the hoistway doors where they are exposed to direct sunlight.
Optimization supplies a plausible, optimized cross-section. It has been demonstrated here that certain geometries can bring about improvements of as much as 35%. One should not, however, overlook potential problems which might result from the joining techniques used here.
1/2004
