Snow transportation by wind action on a wide roof with irregular curved shape placed in urban texture

Transcription

1 Snow transportation by wind action on a wide roof with irregular curved shape placed in urban texture Carmen Elena Teleman Ph.D. Lecturer, Faculty of Construction & Building Services, Technical University Gheorghe Asachi Iasi, Romania carmen_teleman@yahoo. com; teleman@ce.tuiasi.ro Elena Axinte, Ph.D., professor, Faculty of Construction & Building Services, Technical University Gheorghe Asachi Iasi, Romania, e_axinte@ce.tuiasi.ro Georgeta Vasies, M.Bc.in Civil Eng., Ph.D. student Faculty of Construction & Building Services, Technical University Gheorghe Asachi Iasi, Romania, geta_vasies@yahoo.com Summary Structures with wide irregular domed roofs are difficult to design to wind and snow action, the designer facing the lack of information regarding more detailed provisions for the design against these actions in the current codes for practice. A sustainable development of cities starts with a realistic evaluation of the future behaviour of the urban ensemble under the siege of extreme weather manifestations. The paper presents a research program developed in the present days in the Atmospheric Boundary Layer of the Faculty of Construction and Building Services, Technical University Gheorghe Asachi from Iasi. The model of a domed low rise ratio roof was studied for wind and snow effects in atmospheric boundary layer. The building sustaining the roof and all the neighbouring constructions were modelled at a 1:3 scale and tests were run for several wind directions. Snow deposit and drift were studied and the depth of the snow mantle on the roof was arrayed and measured in order to analyze the shape coefficients. In the final stage of the experiments the analysis developed aims to encompass several aspects, from the accuracy of the modelled phenomenon to obtaining more accurate information about the snow drifted deposit for the benefit of a safety structural design. Keywords: boundary layer wind tunnel, wind pressure and snow drifted deposit on domed roofs 1. Introduction Romania is facing, like all European countries abrupt climatic changing and although familiar with severe continental winter weather, unusual snow storms with heavy snow deposits ever more often draw the attention of the design engineers because of the increased safety limits imposed for the security in exploitation of all kind of structures. Increasing the safety limits must not be only a responsible act but also optimise the solutions for design and in this respect the codes for practice give as much assistance that they are prepared to. Any structural designer knows that there are situations and structures for which there might not be enough information for their design against the effects of unusual wind and snow actions. In these cases experimental studies are the only able to offer a clear view on the phenomenon and its impact on the future structure. Studies in atmospheric boundary layer wind tunnels are very appreciated for the extensive data that they provide, helping both the designers and the researchers to understand the effects of these unexpected actions upon the antropic space.

2 The manifestations of wind action in bursts and heavy snow, known as blizzard upon the structures of any kind are difficult to study and describe at natural scale and even more to simulate in laboratory. Snow simulation is still a daring endeavour, as its density varies widely, depending on temperature, air pressure and wind speed which transports it but the usual studies in wind tunnel cannot reproduce the chemical-physical mechanisms responsible for these modifications. The concentration on the identification of the type and proportion of particles of snow drops that are transported by wind on the surface of the buildings is then very important for a correct estimation of the snow deposit. A wide and various number of studies were dedicated for these complex phenomena in the last more that 4 years and the variety of approaches show themselves the almost discouraging number of impediments in the simulation in the laboratory. In order to obtain step by step results, priority simulation criteria must be imposed, considering the specific effect that is currently studied. The study presented herein gives a helping hand to the structural engineer interested in both the shape of the snow agglomerations on the roof and on the estimation of their depth, having in mind previous severe accidents due to exceptional snow falls (crash of the domed roof of the National Expo Center in Bucharest in 1963 where from measurements it was found that locally, the snow deposits exceeded 2 3 m even more than 3.5 m height). Wind increases the density of the snow and hardens the cohesion between particles, remodeling the shape and the distributions of the deposits on the roof. Wind creates distinct areas on the snow deposits, the particles being transported on the surface of the roof from points characterized by increased turbulence. The dynamics of local vectors of the gradients of wind instantaneous speed are responsible of both transportation and deposits of the snow. A critical wind speed of 1 km/h at 2 cm above the ground would in general determine snow transportation, the volume of the snow transported increasing cubic with the wind speed. Wind may produce blizzards at speeds exceeding m/s and over15 m/s it is reported that snow concentration in transport exceeds 5g/minute x cm 2 [1]. In Romania maximum wind speeds in blizzards are in general 24 m/s 3 m/s, but most likely, the mean speed of wind in blizzard is between 11 m/s and 17 m/s, important snow agglomerations being observed from 6 m/s up, on periods of about hours. 2. Experimental studies in wind tunnel 2.1 The object for study The studies upon wind action and snow drift presented here find their opportunity in the unusual large dimensions and irregular shape of a domed roof covering a complex of buildings and a plaza situated in the middle of an important urban centre in Iasi, a big city of Romania. The roof of about 1 m diameter is placed lower that the surrounding very high buildings, at about 2 m and the whole area has a pronounced variation of the ground level. The roof angle of inclination is under 3 and radial northern lights trace concentric circles placed at uneven intervals on its surface. The height of these vertical northern lights is 1.5 m [7]. Neither the codes for snow loads [2] nor those for wind loads [3] are able to offer the correspondent values for a safety design of this roof so these atypical situations must be studied in a wind tunnel. The atmospheric boundary layer wind tunnel SECO 2 is with open circuit, a transversal cross section of 1.4 m x 1.4 m and its experimental length is 8.8 m. The maximum speed developed may be 18 m/s but in usual cases the model of the gradient speed reaches 9 1 m/s at the top part of the tunnel. The building and a sufficiently large area surrounding it were modeled at 1:3 scale and placed on the turntable. The local dynamic wind speed is measured with hot wire anemometry system IFA3, (T.S.I., SUA) and the wind dynamic local pressures are measured at pressure taps on the surface of the model of the building and transformed in electric signals which are captured by a ZOC 17 (Zero Calibrate and Operate) pressure module (SCANIVALVE S.U.A.) and transferred to a acquisition board NI- DAQ, National Instruments, S. U. A. Specific software processes data obtained via measurements.

3 2.2 Simulation of the wind action in the atmospheric boundary layer The accuracy of the information obtained via experiments is obtained only if the model of the wind flow in the boundary layer and the geometry are scaled to the prototype. A boundary layer specific for the urban environmental conditions was artificially developed in the tunnel and the characteristics of the simulated turbulent flow are shown (the profile on the height of mean longitudinal wind speed, turbulence intensity, standard deviation) in figure 1. a. b. c. Fig.1. Dynamic characteristics of the turbulence of the urban boundary layer: a.-mean wind speed; b.-standard deviation, c.- turbulence intensity At the reference height of the model (height of the roof), the turbulence intensity u / uz (%), is about 2%...24%, specific for the urban area and the mean wind speed has a power law coefficient α=,33 (the speed at the top of the ceiling in tunnel was fixed around 9m/s). A blockage area of 5...1%, in the tunnel cross section was set and verified not to be exceeded by the dimensions of the model in the transversal section (1...2 cm 2 ) Acquisition of local dynamic pressures A number of 125 pressure taps were placed on the cupola, mainly along radial directions. In the figure 2 left, the distribution of the wind pressure taps may be observed on the model and in the right side pressure taps on a sector of the roof extracted for the further analyses. The pressure taps correspond to those for which the mean values of wind pressure are presented in graphs from figure 3 a and b. The model was studied in 12 wind attack angles each with a 3 increment. The values of the local instantaneous pressure are registered for a 1 minute with a frequency of 1 Hz, a number of 6. samples per pressure tap resulting. The time scale sets a 1 minute in the tunnel equivalent with 3 minute in nature. The frequency at natural scale is then 2 sec/value of dynamic pressure. The analysis of all the mean values of the pressure coefficients showed that regardless of the wind direction they are negative, the roof being under general suction forces, explained by the low inclination and the presence of sharp edges of the northern lights which did not allow to the flow to reattach to the surface. In the same time, from the observations it resulted also that in the cases when the flow acts first on another building big enough, the roof is sheltered, the wake generating low effects of delta wings locally increasing the suction. Still, the most important observation is that the bigger values of mean and extreme values of pressure negative coefficients are not in centre of the roof and not at the edges, but in between, where the angle of inclination changes and the skylights are more numerous, the same place where the drifted snow is expected to deposit most. In the graphs, the variation of the mean pressure coefficients with the wind attack angle is presented, in 2 groups of six angles each. The points in which the local values are presented are placed radial, starting from the centre of cupola (the position and identification of the pressure taps are presented in figure 2 (right).

4 Fig. 2. View of the model of the building with the domed roof for the wind attack angle (N) (left) and the position of the pressure taps on the analysed sector (right) [7] 1 Fig. 3. a Variation of the mean negative pressure coefficients with the wind attack angle,3,6,9,12,15, on the pressure taps 4, 5, 6, 117, 11, 119, 12, 39, 42 Fig. 3 b Variation of the mean negative pressure coefficients with the wind attack angle 18,21,24,27,3,33, on the pressure taps 4, 5, 6, 117, 11, 119, 12, 39, 42

5 3. Study of the snow drifted on the curved roof 3.1 Simulation of the snow drifts in the atmospheric boundary layer wind tunnel Simulation of snow drifted deposits with glass micro-balls used as reflecting material in road marking signs is satisfying. Snow is a granular material in dispersion with various densities, corresponding to air humidity, temperature and wind speed and these shifting values are stressing the design engineer. The snow particle diameter varies from.1 mm at C to.5 mm at very low temperatures and under the erosion effect of wind blow. Under the effect of sun heating, when inclination angle is under 5, snow melts at the surface in a very tough layer, resisting for a longer period. From the total mass of snow, more than 8% is formed by the particles transported in blizzard, responsible of snow drifted deposits. Simulation of the snow agglomeration on roofs must respect some basic criteria, one being the match of the geometric shape. Other important simulation criteria of the transportation rate and volume refer to Jensen criteria of roughness, Reynolds number, Froude number, time, length and density scales between the prototype and the model. In the case studied, the particles that simulate the snow are heavy and studies showed up to now that with the exception of Froude number, the other criteria are respected once the wind flow in the tunnel is scale modeled [4]. Froude number is almost impossible to be respected because of the necessity that wind speed has a threshold which must be exceeded in order to move the particles of the simulating material. Some authors consider that this number may be reduced as long as the particles trajectories are far smaller that the dimensions of the physical models, so a realistic snow drift simulation must be on larger geometric scales. The scale of simulation of this study remained 1:3 because of the restriction imposed by the cross section of the wind tunnel and because of the necessity of preservation of all the respected criteria of wind flow simulation. 3.2 Experimental conditions The simulation of snow drift in the Laboratory of Aerodynamics is made with the help of a remote controlled equipment able to release a flux under control of granulated material that simulates the snow particles of snow (figure 4). The equipment is placed in the tunnel normally to the air flow and in front of the turntable, and the particles of simulated snow are dragged by the speed of the wind, reproducing the trajectory of the natural snow drops. Fig. 4. Equipment used for the physical model of the snow blowing (left). Model of the roof and surroundings with snowdrift (right) The loose material used for the snow simulation is a combination of granulations of glass balls with diameters of.2.6 mm about 9% of its quantity and with a density P = 1.4 g/cm 3. The wind speed during the experiment was set considering studies priori developed in the same tunnel with the same granular material [1], during these experiments the registered values at the

6 top of the tunnel section were about 6 m/s. In order to obtain a saturation of the deposits of the blowing particles and a reference scaled period of time, the tests were run with the same quantity of material and lasted the same, each time about 5 minutes (minimum 8 hours at natural scale). A preliminary experimental stage was developed for the determination of the depth of the deposit on the ground, with no obstacles. The conditions of testing were set identically with the general conditions: the same wind speed and the same period of time (same quantity of material). The surface on which the particles simulating the blowing snow were transported is a flat polystyrene, the same material used to model the roof marked with 17 markers placed at 3 cm one from the other, covering a surface of more than 1 cm 2 (about 1% of the surface of the roof in plane). During the experiment the averaged longitudinal component of the wind speed was measured, u z in two points in the middle of the flow, on the height of the tunnel; one height z1 was defined in accordance with the roughness length z and the other z 2 is the reference height, which is the height of the curved roof of the model, at 7.4 cm from the floor. The measured values of the wind speed in these two points were m/s at the reference height and m/s on the height of the aerodynamic roughness. The roughness length or the aerodynamic roughness is a constant used in the logarithmic law of the wind speed profile and in urban texture is about.5 1 m. As the measurement takes into account the thickness of the polystyrene plate, of 2.2 cm, the height was scaled to 2.5 cm. Based on the velocities determined for z cm and z cm, the relationship (1) for the determination of z was used and (2) for the friction velocity: z2 ln z1 uz1 ln z2 u z2 u z1 u z exp (1) Fig. 5. Drifted particles deposited on a flat surface (left); array of markers in the layer of glass balls (right). Left- Pitot tube placed at the 2.5 cm from the ground floor and the wind speed transducer which collects instantaneous values and processes data online k u* uz (2) ln z z In the relationship (2) k is the roughness factor, which for urban terrain is The result from (1) gives a good match because for z a value of.2cm shows that the value in nature is 6 cm, that is a standard value; the result from (2) gives a value of.1.11 m/s, which is situated in the interval of.9 m/s reported in literature [5], for wind speeds of 1 2 m/s. The ratio:

7 u z u z respects the criteria of simulation of the particles in the saltation mode in the absence of any other factors of pressure, temperature and humidity of the air, that is the ratio between the particle drift speed and the threshold friction velocity must be bigger than 1 [6]: U u f * t 1 (3) (4) 4. Results 4.1 Data obtained from measurements The tests were run in concordance with 5 observed situations, when wind action generates the possibility of a more consistent deposit of snow on the roof. In the figure below a pattern is presented of the drifted snow deposit on parts of the roof. In order to evaluate the snow deposits, markers were placed on the roof in the close vicinity of the pressure taps and at the base of the northern lights, in order to obtain the depth of the deposit fight in their front. Markers were read at the end of each measurement giving sets of values of the deposit of the glass micro-balls (figure 6). Fig.6. Measurements read on markers (in mm) with a magnifying glass: left- for the wind attack angle = 6 ; right- for the wind attack angle = Data analysis In order to use the data obtained via measurements in the evaluation of the snow drifted deposits on the roof we tried to find a method for validation of these values. First, a sector of the roof was chosen, which put itself in evidence during the whole research program as being strongly affected by wind action and snow deposits. The depths of the snow deposit read on markers presented in figure 7 were processed for the particular sector of cupola.

8 Table 1. Coefficient μ of variation of the snow deposit on the cupola Wind attack angle Mean value, x Standard deviation, s The coefficients of variation between the deposit on the roof and the deposit on the ground were determined for the sector of cupola: x i i (5) S d where: - i - coefficient of variation of the depth of the snow deposit; - x i - measured depth on the markers; S - averaged value of the depth of the layer on the ground. - d Table 2. Coefficient μ of variation of the snow deposit on the cupola Pressure tap mean µ(6º) 1.567; r. m. s.3956 mean µ(9º) ; r. m. s An approach based on statistical interpretation of data is used afterwards, in order to give more information about a representative mean value of the deposit of the drifted snow; the comparison between the representative mean values of the snow deposits on the roof and on the ground. The mean values were determined using the Student t distribution, based on similar analysis in other studies. The estimation of a mean value is based on the variable t determined for 95% reliability and for a number of freedom degrees, which are the number of measurements on the sector of cupola (13 points) and on the ground (17 points), so: x t (6) s N where : t - Student t variable; x - mean of the sampled values x 1,x 2,x i x n : N xi i x 1 (7) N

9 N s - standard deviation (root mean square) : 2 2 s 1 x i x (8) N 1 N - the number of samples, in this case the number of measurements on the roof sector; - estimated mean value of the snow deposit. The relationship (6) is used to determine the estimated mean values of the snow deposit on the sector of the roof presented in table 1 and the estimated mean values of the deposit on the ground. The variable t was determined for a 95% reliability percentage. The values are presented in table 3. Table 3. Values of t variable and of the expected mean depth of the snow deposit On the cupola (Wind attack angle) i1 6 t,5, mm mm t,5, mm On the ground 11 5 Conclusions 5.1 Discussions The study was dedicated to the snow drift on the cupola in the presence of wind. A vertical section through the roof reveals a number of multi span roof sheds; for the sector of cupola that was analyzed, a radius of 96 m is formed of radial stripes of variable widths: from 4.39 m the lowest to m the upper stripe, the diameter of the central circle being m (fig 3). Every stripe is divided in two, one following the general inclination of the cupola; the other follows the ridge of the northern light, generating a valley of variable widths, responsible for snow agglomerations with, or without wind action. The general inclination of the roof being 13.4, first conclusion is that the maximum coefficients of un-uniformity of the snow deposit presented in table 2 match the recommended values for the shape coefficient 2 from the code for design [2], in the case of roofs with angles of inclination under 3 (the case of snow being a variable action in the design to ultimate limit states). Considering the situations of a totally uneven snow deposit, the recommendations in [3][8] even if not very appropriate for the roof studied, leave to the structural engineer the decision to adopt a minimum value from several geometric ratios concerning the dimensions of local irregularities on the roof. The local irregularities are determined by the presence of the northern lights, the 1.5 m height windows being vertically placed at the limits of the concentric stripes; an evaluation at this level based on the specifications referred to, give an important conclusion that it might happen the situations when the northern lights will be snowed up, the snow deposit literally reaching 2..3 times the height of the vertical window and this is the case observed already during the experiments. The attempts made for statistic interpretations of the deposits on the roof bore in mind the fact that the data are random and that they may be either not representative to a certain degree and in the same time affected by a lack of precision in the simulation. Still, it is rather spectacular and encouraging finding that the Student t distribution gave expected values of the mean of the snow drifted deposit that may be compared in between. The mean expected value of the depth of the deposit on the flat plate of polystyrene placed at the limit or aerodynamic roughness is 5.19 mm between the values obtained on the roof, 4.8 mm for 6 and 5.4 mm for 9 wind angle attack; the differences may be tributary to the inclination of the roof, which generally diminishes the height of the deposit but on the other hand, the variation of the local geometry of this roof is the reason of the agglomerated deposits. The wind simulation and afterwards processed data [7] gives credit to a good simulation of the boundary layer so the local turbulence is supposed to be the reason of the variable values of the deposits on the roof. The explanation is that, even the mean values are almost entirely of suction, the maximum fluctuating values of the local coefficients are in the rage of about ( and ). Local burst may increase or decrease the deposit in the valleys between northern lights. It is important not to forget that, from all data registered, the analysis uses the worst of the situations, considering the environment and the wind attack angle.

10 5.2 Final remarks The series of experiments described help to a better understanding of the way the material used to simulate the snow drift works and the key role played by all the factors implied in the simulation. Modelling in wind tunnel at reduced scale the structure and its surroundings gave better idea about the influence exerted by the other buildings and by the irregularities of the terrain on the studied roof and helped decide upon the worst situations the designer should take into account. The interpretation of the results although probably at this stage does not take into account many other simulation aspects and limits, shows that this is a good direction for the study of snow action on buildings in our laboratory. Further analyses will enhance the accuracy of these experiments. The design to the combination of wind and snow action of this roof must take into account all the situations imposed by the codes for practice but also observed in laboratory, because only there the simulation took into account all the factors of influence. It is ever more obvious that the designing engineer must nowadays cooperate with the researchers for a better understanding the complex phenomena that affects the service life of our urban habitat and provide the necessary security of their exploitation. 6 References [1] Florescu E. C., Modelarea parametrilor climatici in vederea obtimizarii elementelor geometrice ale profilului transversal al drumului in regiuni cu ierni aspre, Teza de doctorat, Universitatea Tehnica Gheorghe Asachi Iasi, Romania, 21 [2] SR EN Actiuni asupra structurilor. Partea 1-4: Actiuni generale - Actiuni ale vantului [3] SR EN Actiuni asupra structurilor. Partea 1-3: Actiuni generale. Incarcari date de zapada [4] Iversen J. D., Smal-scale modeling of snow-drift phenomena, Congress Wind Tunnel Modeling for Civil Engineerin, Gaithersburg, Maryland, U.S.A 1982, pp [5] Thiis T. K., Large-scale measurements of snowdrifts around flat-roofed and single-pitchroofed buildings, Cold Regions Sience and Technology 3 (1999) pp [6] Leitl B., Schatzmann M., Baur T., Koenig-Laglo G., Physical modeling of snow drift and wind pressure distribution at the proposed german antarctic station NEUMAYER III, 25 th International Conference on Offshore Mechanics and Arctic Engineering, June 4-9, Hamburg, Germany, 26 [7] Teleman E. C., Axinte, E., Studiu in tunel aerodinamic privind actiunea vantului si a zapezii asupra acoperisului cupola din cadrul ansamblului PALAS Iasi, research contract nr. 343, CCTT Polytech - Universitatea Tehnica Gheorghe Asachi Iasi, Romania, 21 [8] Cod de proiectare. Evaluarea zapezii asupra constructiilor. (Revizuire CR Comentarii. Recomandari de proiectare si exemple de calcul.) Redactarea I, 21