Laboratory Study on a Duct-Ventilation Roadbed of the Qinghai-Tibet Railway

Laboratory study on a duct-ventilation roadbed of the Qinghai-Tibet railway

F.J. Niu, G.D. Cheng & Y.M. Lai State Key Laboratory of Frozen Soil Engineering, Cold and Arid Regions Environmental and Engineering Research Institute, CAS, Lanzhou, People’s Republic of China

ABSTRACT: The efficiency of using ventilation ducts in the roadbed for the Qinghai-Tibet railway was studied with two large-scale laboratory experiments. The results confirmed that it is easier to freeze the pore-water within the railway roadbed when ventilation ducts are used, and that the ventilated roadbed protects existing frozen ground from thawing and can even extend the frozen zone. However, the temperature distributions along a section parallel to the ventilation ducts are uneven, and this also varies with different distances from the duct, so that the temperature is uneven in three dimensions. As a consequence, uneven heave and thawing deformations might be caused in real constructions. The conclusions have led to recommendations for the design of a field experimental project using a ventilated roadbed along the Qinghai-Tibet railway in the Beiluhe region, where huge underground ice layers are very common.

1 INTRODUCTION roadbed, constructing a railway-bed on insulating layers, making a reserve by increasing the roadbed width The Qinghai-Tibet railway is the longest railway at high and height, and ventilating the roadbed. Among these altitude in the world. It runs from Xi’ning, the capital of methods, the first three are passive to prevent the Qinghai province, to Lasha, the capital of the Tibet degradation, needing too much cost for construction autonomous district, China. The total length of the line and maintenance. These methods are also aimed at is 1,959 km, of which 841 km has been completed overcoming the consequences of existing permafrost (between Xi’ning and Golmud) and the remaining sec- degradation (Kondratjev 1996). Ventilating the roadbed tion (from Golmud and Lasha) is 1,118 km long and is an active method to remove heat from the ambient crosses the hinterland of the Qinghai-Tibet high soil, so that the primary permafrost table can be kept plateau. Of this section, 630 km’s of railway line will stable or even raised. cross permafrost with mean annual air temperature Though roads and tunnels have been constructed on of 7 to 2°C. The high altitude of the plateau means the high plateau and relative temperature calculations that there is a large-area of permafrost, which causes have been done (Li & Wu 1998, Lai et al. 1999), there many engineering difficulties to railway construction. is no experience of constructing railways on such high These problems include subsidence, frost heave, under- altitude land, especially when the railway crosses ground ice and debris flow caused by melting ice. ice-rich or high-temperature permafrost regions. To Under the influence of both global warming trends and obtain basic knowledge and initial data of ventilation engineering construction, thaw settlement might occur, of the roadbed and due to the tight time schedule of leading to destabilization of the railway roadbed and field-testing projects, laboratory studies have been failure of the structure. The most hazardous damage is pursued in advance of fieldwork. The main object of due to degradation of the permafrost and thawing set- the experiments is to compare the temperature distri- tlement of the roadbed. In fact, most failures of railway bution in the roadbed with and without ventilation roadbeds in permafrost regions are due to the degrada- ducts, and to analyze the ventilation effect. tion of permafrost caused by engineering influences. The ventilation method is widely used in the con- For example, about 70% of roadbed deformations struction of garages, warehouses and storage tanks in along the Baikal-Amur railway are due to the degrada- permafrost. In the past, two methods of construction tion of permafrost (Belozerov 1993). have been adopted. One is to install open-ended ducts Presently, two engineering principles are applied in in the pad or roadbed, oriented in the direction of the preventing the deformations caused by the degradation. prevailing winds (Nixon 1978). Another one is based The first one is to protect the frozen soil, and the other on the chimney effect, to induce air movement in one is to pre-thaw the frozen soil (Cheng 2001). stacks connected to the ventilation ducts (Tobiasson Currently, the Qinghai-Tibet railway is mainly required 1973). As it is very windy on the Qinghai-Tibet high to be constructed under the first principle. Proposed plateau, the roadbed with an open-ended ventilation methods include the following: using a crushed stone duct is studied in this paper.

815 2 VENTILATION ROADBED MODEL Air temperature, soil temperature, soil deformation and wind speed were monitored in the experiments, The experiment consisted of five parts. They are the with thermal sensors, displacement sensors and model box, the cooling system, the wind system, the anemometer respectively. Data were collected every model soil body and the monitoring system. Each of 20 minutes by a DATATAKER 500, and then trans- them is introduced below. ferred to a computer for subsequent analysis. The inner dimension of the model box is 8.0 m long, 1.84 m wide and 2.7 m high, and it is constructed from a 10 cm thick insulation plate. The box was reinforced 3 RESULT OF MODEL EXPERIMENTS with a steel frame on the outside, so that the wall can resist the vertical and lateral soil pressure during soil 3.1 Experiment No. 1 compaction. The cooling system consists of a refrigerating In experiment No. 1, compacted loess and crushed machine and cooling pipes. Cooling media is com- stone were used. The soil model is 1.5 m thick, 5.8 m pressed and delivered to fans in the model box. The air wide at the bottom, 2.8 m wide at the top and 1.84 m temperature in the box can be controlled automati- long, and a 0.5 m thick crushed stone covers it. The cally between 30 and 40°C. model is frozen under 7 to 5°C for 480 hours, and The wind system consists of cooling fans, heaters, then thawed for 240 hours. Two temperature monitor- a wind-speed-controlling sheet, a wind-direction- ing transects, named A and B, are put in the model. adjusting shutter and an air-returning tunnel. The Section A is located along the central section of the air-returning tunnel is used to recycle air in these body and section B is 0.50 m away from the center. experiments. Wind speed in the box can be set freely in the required speed range during the experiments, and 3.2 Experiment No. 2 the prevailing wind direction is set parallel to the long side of the box and is vertical to the extending direc- The model soil body in experiment No. 2 is the same tion of the model roadbed. In the experiments, the air size as in No. 1, while a 4.8 m long duct, with an temperature is about 7 to 5°C and the wind speed inner-diameter of 0.28 m was put in the soil body at a is 2 m/s during the freezing period and 18–20°C during height of 0.5 m above ground level. The two monitor- the thawing period. ing sections are the same as in experiment No. 1. The As the purpose of the experiments is to determine soil is also frozen for 480 hours and then thawed for temperature distributions of the roadbeds with and 240 hours. without ventilation ducts, and extracting massive Five deformation monitoring sensors are placed on insitu soil samples is difficult, the soil mass used in the surface of the soil models. According to the mon- the experiments is loess from Lanzhou. Physical itored data, the maximum frost heave is only 3 mm. parameters of the soil are tested for use in further calculations, and these are listed in Table 1. Two types of sample bodies are constructed in the experiments, one 3.3 Experiment results is a normal railway roadbed and the other is a ventilation roadbed. Each of the models is 2.0 m thick, 5.8 m 3.3.1 Temperature changing process wide at the bottom, 2.2 m wide at the top and 1.84 m A total of 60 thermal sensors are used in the experi- long, which is nearly 1/4 scale of the real roadbed of ment to monitor temperature, four of them are located the Qinghai-Tibet railway. at critical positions in the soil body (Fig. 1). The sensors at positions P1 and P2 are used to represent the effect at the initial ground surface (the bottom of the Table 1. Physical parameters of the experimental soil. soil body is insulated in the experiments); the sensor Parameter Ranges of values Unit 200 Dry density 1.46–1.50 g/cm3 Crushed stone 150 Water content 12.9–13.6 % P4 Thermal 0.833(23°C)–0.947(6°C) W/m °C 100 Loess conductivity P3 Ventilation duct Thermal 1.47(23°C)–2.0(6°C) 103m2/h Height (cm) 50 P1 P2 diffusivity 0 Specific heat 1.27(23°C)–0.99(6°C) kJ/kg °C 0 100 200 300 400 500 600 Specific heat 2187(23°C)–1705(6°C) kJ/m3 °C Width (cm) capacity Figure 1. Four typical temperature-monitoring positions.

816 at P3 indicates the temperature at the center of the soil that ventilation duct can cool the roadbed soil body body; and P4 measures the temperature at the surface down more quickly than the normal roadbed structure of the soil body below the crushed stone layer. The does. At the beginning of the freezing time, the cooling soil temperatures are shown in Figure 2 at the four velocity is high and then it decreases slowly. sensor positions for a total 740 hours. The surface temperature of the model soil in the two Figure 2a clearly indicates that temperatures dropped experiments is nearly 2°C different at P4 at the begin- from 12.4°C and 12.1°C to 12.4°C and 12.3°C, ning of the experiments, but soon they tend towards the respectively at positions P1 and P2 after 120 hours of same value. After 20 days of freezing, the differential freezing in experiment No. 1. While in experiment value between temperatures at P4 in the two experi- No. 2 (Fig. 2b), the changes are from 15.5°C to ments is only 0.11°C. This shows that the surface tem- 10.6°C and 15.6°C to 11.3°C, respectively. perature is mainly controlled by the air temperature and The values show the differences between the chang- is not related to the presence of a ventilation duct. ing ranges in the two experiments. In addition, although In general, the different freezing processes of the the initial temperature values of the soil in experiment soils in the two experiments show that it is much easier No. 2 are about 3.0°C higher than those of experiment to lower soil temperatures in a ventilated roadbed than No. 1, they are more than 1.0°C lower than these in No. a normal roadbed, especially during the earliest peri- 1 only after 120 hours of freezing. Another phenome- ods of freezing time. Therefore a ventilated roadbed non is that the differential temperature value between could be efficient in protecting the thermal state of the the two experiments at the same position of P1 and P2 original frozen ground and even consume heat from increases with time during the first 240 hours (3.1°C at the ground. P1 and 2.5°C at P2) and then decreases. After another The temperatures at the four positions in the two 240 hours, the difference between the two experiments experiments are also different during the thawing at P1 and P2 is 2.4°C and 2.1°C, respectively. period. Although the temperatures of the soil bodies The difference between the two experiments reaches are different at the beginning of thawing, after 240 3.9°C after 120 hours of freezing at position P3, and hours of thawing, the temperatures in the two experi- even extends to 4.0°C after the next 120 hours. This ments are nearly the same. This indicates that if venti- means that the temperature difference is more distinct, lated roadbeds were to be adopted for construction of and the temperature changes at P1, P2 and P3 indicated the Qinghai-Tibet railway, the ducts should be closed in warm seasons, or some other measures should be used to prevent thawing from occurring quicker than 20 Air Temperature for an unventilated roadbed. 16 P4 12 3.3.2 Temperature field distribution P2 As temperature distribution of the two sections of 8 P1 experiment No. 1 is nearly similar, only one of them 4 has been compared with the two sections of experiment P3 0 No. 2. Figure 3 shows the initial temperature distribu- Temperature ( º C) 0100 200 300 400 500 600 700 800 tions in the soil body of the two experiments. Although -4 temperatures in the soil bodies are uneven (12–17°C), -8 the general temperature conditions of the two can be (a) Time (h) considered to be more or less the same. This means that 20 the experiments are started under similar conditions, Air Temperature except that the temperature is difference at the bottom 16 of the two soil bodies. The temperature at the bottom of 12 the soil body in experiment No. 1 is 12.1–12.4°C, while P4 that of experiment of No. 2 is 15.4–15.6°C. 8 P2 P1 Figure 4 shows temperature distributions of the 4 three sections after 240 hours of freezing. Figure 4a P3

Temperature ( º C) 0 indicates that the freezing front (0°C line) in the 0100 200 300 400 500 600 700 800 model body of experiment No. 1 extends to the depth -4 of 63 cm, while that of experiment No. 2 reaches -8 76 cm below the top surface (Fig. 4b). In section A of (b) Time (h) experiment No. 2, the whole part above the ventilation Figure 2. Temperature curves at the four sensor positions duct and about 2/3 of the whole soil body would have in the two experiments (a. experiment No. 1, b. experiment been frozen (Fig. 4c). The temperature at the bottom No. 2). of the soil body of No. 1 decreases to 9.6–9.7°C while

817 200 No. 1-B 200 No. 1-B 150 150

100 100 Height (cm) Height (cm) 50 50

0 0 0 50 100 150 200 250 300 350 400 450 500 550 050 100 150200 250300 350400 450500 550 (a) Width (cm) (a) Width (cm)

200 200 No. 2-B No. 2-B 150 150

100 100 Height (cm) Height (cm) 50 50

0 0 0 50 100 150 200 250 300 350 400 450 500 550 050 100 150200 250300 350400 450500 550 (b) Width (cm) (b) Width (cm)

200 200 No. 2-A No. 2-A 150 150

100 100 Ventilation duct Ventilation duct Height (cm)

Height (cm) 50 50

0 0 0 50 100 150 200 250 300 350 400 450 500 550 050 100 150200 250300 350400 450500 550 Width (cm) (c) Width (cm) (c)

Figure 3. Temperature distributions along sections of the Figure 4. Temperature distributions along sections of the roadbed models (t 0 h, unit: °C, a. section B in experi- roadbed models (t 240 h, unit: °C, a. section B in experiment No. 1, b. section B in experiment No. 2, c. section A ment No. 1, b. section B in experiment No. 2, c. section A in experiment No. 2). in experiment No. 2). that of No. 2 decreases to 6.5–7.2°C: the difference is this is that the thawing starts at a different temperature clearly apparent. On the other hand, Figure 4 also in the two experiments. In section A of experiment indicates that the uneven phenomena of temperature No. 2 (Fig. 6c), the general temperature is higher than distributions in experiment No. 2 is more distinct than that in section B of both experiments No. 1 and No. 2. those in experiment No. 1, which means that the most This means that the open-ended ventilation ducts possible influential factors are the introduction of the should be closed in warm seasons, otherwise, additional duct, as well as the wind direction. thawing settlement might occur, leading to failure. Figure 5 shows the temperature distributions at the The model tests of a single ventilation duct indicate end of the freezing period (after freezing for 480 that a ventilated roadbed is efficient in protecting hours). The three figures indicate that in experiment frozen ground from thawing. Based on these results, No. 1, the freezing front reached 76 cm below the top a 400-meter-long field project has been designed for surface (Fig. 5a), and 92 cm in section B of experi- the Beiluhe region of the Qinghai-Tibet plateau. The ment No. 2 (Fig. 5b). After 480 hours of freezing, the climatic character of this region shows that the freezing temperature at the bottom of the soil body of No. 1 time is 8 months (from September to April). During decreases to 6.1–6.3°C and that of No. 2 decreases to this period, the dominant wind direction is NW, and the 3.7–4.2°C. Meanwhile, nearly the whole area of sec- annual-mean wind speed is 4.1 m/s. These conditions tion A of the model body in experiment No. 2 has are similar to those of the laboratory experiment. been frozen (Fig. 5c). This figure provides important Therefore, the results obtained from the experiments information that the ventilated roadbed is much more are valuable to the field-testing project, which is cur- efficient than the normal roadbed in keeping the soil rently under construction. frozen. The results of the experiments revealed the tem- Figure 6 shows the temperature distributions of the perature distributions throughout the model roadbed. model soil bodies after 240 hours of thawing. In section However, the experiments were carried out with only B (Figs 6a, b), the temperatures in the soil of experiment a single wind direction with stable temperature, No. 2 are generally lower than in No. 1. The reason for whereas in a real project, the circumstances would be

818 200 No. 1-B 200 150 150

100 100 Height (cm)

Height (cm) 50 50

0 0 050 100150 200 250 300 350400 450 500 550 050 100 150 200 250 300 350 400 450 500 550 (a) Width (cm) (a) Width (cm)

200 No. 2-B 200

150 150

100 100 Height (cm) 50 Height (cm) 50 0 050 100150 200 250300 350 400 450 500 550 0 050 100 150 200 250 300 350 400 450 500 550 (b) Width (cm) (b) Width (cm) 200 No. 2-A 200 150 150 100 Ventilation duct 100 Ventilation duct

Height (cm) 50 Height (cm) 50 0 0 50 100 150 200250 300 350 400 450 500 550 0 Width (cm) 050 100 150 200 250 300 350 400 450 500 550 (c) (c) Width (cm)

Figure 5. Temperature distributions along sections of the Figure 6. Temperature distributions along sections of the roadbed models (t 480 h, unit: °C, a. section B in experi- roadbed models (t 720 h, unit: °C, a. section B in experiment No. 1, b. section B in experiment No. 2, c. section A ment No. 1, b. section B in experiment No. 2, c. section A in experiment No. 2). in experiment No. 2). much more complex. In future experiments, a variable temperature distribution. Therefore, uneven frost temperature should be considered so that the real con- heave and thawing deformation must be consid- ditions could be modeled more closely. ered in real projects if the soil is frost-susceptible. (3) During periods of high temperatures (or in warm seasons), the open-ended ventilation ducts should 4 CONCLUSIONS be closed, otherwise the roadbed will tend to thaw and additional thawing settlement might occur. (1) The temperature distributions in the experiments with and without a ventilation duct indicate that the soil body in a ventilated roadbed is easier to ACKNOWLEDGEMENTS freeze than soil in an unventilated roadbed. After 480 hours of freezing, the temperature at the bot- This study was partly supported by the Chinese Natural tom of the soil body in the ventilated roadbed is Science Foundation (No. 50008016), the Foundation reduced by 11.3–11.8°C, while in the unventilated of “Hundred Intelligent People Plan” (to Y.M. Lai) roadbed, it is reduced only by 5.9–6.3°C. Therefore, of CAS, and the Knowledge Innovation Foundation of the ventilated roadbed is efficient in protecting CAS (No. KZCX1-SW-04). permafrost under a railway roadbed and can even extend it. Such a result is also expected during construction of Qinghai-Tibet railway if ventilated REFERENCES roadbeds are adopted. Belozerov, A.I. 1993. Roadbed deformations at the Baikal- (2) The temperature distribution in a ventilated Amur Railway. Transport Construction 5–6: 11–14. roadbed along a section parallel to the ventilation Cheng, G.D. 2001. Linearity engineering in permafrost duct is uneven under cooling wind blowing in a areas. Annual Report of State Key Lab. Of Frozen Soil single-direction. Moreover, sections at a different Engineering, CAS 10: 95–100. Lanzhou. in Chinese distance from the duct also show a very different with English Abstract.

819 Kondratjev, V.G. 1996. Strengthening railroad roadbed bases East of Tibet Plateau, China. Cold Regions Science constructed on icy permafrost soils. In Proceedings of and Technology 28: 183–188. Eighth International Conference on Cold Regions Nixon, J.F. 1978. Geothermal aspects of ventilated pad Engineering: 688–699. Alaska: Fairbanks. design. In Proceedings of the Third International Lai, Y.M., Wu, Z.W., Zhu, Y.L. & Zhu, L.N. 1999. Conference on Permafrost: 841–846. Edmonton: Nonlinear analysis for the coupled problem of temper- Alberta. ature and seepage field in cold regions tunnels. Cold Tobiasson, W. 1973. Performance of the Thule hangar soil Regions Science and Technology 29: 89–96. cooling systems. In North American Contribution. Li, D.Q. & Wu, Z.W. 1998. Heat stability analysis of Second International Permafrost Conference, Yakutsk: roadbed on the degrading permafrost district in the 752–758.

820