Evaluation of Critical Parameters to Improve Slope Drainage System

+is study focuses on identifying and evaluating critical parameters of various drainage con0gurations, arrangement, and 0lter which a1ect the e2ciency of water draining system in slopes.+ere are a total of seven experiments with di1erent types of homogeneous soil, drainage envelope, 0ltermaterial, and quantity of pipes performed utilizing amodel box with a dimension of 0.8m× 0.8m× 0.6m.+e pipes were orientated at 5 degrees from the horizontal. Rainfall event was introduced via a rainfall simulator with rainfall intensity of 434.1 mm/h. From the experiments performed, the expected outcomes when utilizing double pipes and geotextile as envelope 0lter were veri0ed in this study. +e results obtained from these experiments were reviewed and compared with Chapter 14 “Subsurface Drainage Systems” of DID’s Irrigation and Agricultural Drainage Manual of Malaysia and the European standard. It is recommended that the pipe installed in the slope could be wrapped with geotextile and in tandem with application of granular 0lter to minimize clogging without a1ecting the water discharge rate. Terzaghi’s 0lter criteria could be followed closely when deciding on new materials to act as aggregate 0lter. A caging system could be introduced as it could maintain the integrity of the drainage system and could ease installation.


Introduction
Malaysia is located in the tropics where heavy rainfall and thorough in situ chemical and mechanical weathering result in the development of deep residual soil pro les.Slope failures are a ected rapidly by rising groundwater level and rainfall in ltration due to the frequent high-intensity tropical rainfalls.Installation of horizontal drains in slope is one of the common methods used by engineers to lower the amount of excess water in slopes.Horizontal drains could be de ned as holes drilled into a slope and cased with perforated metal or slotted plastic linearly to drain out groundwater [1,2].Horizontal drains have been used successfully to improve slope stability [2][3][4][5][6][7].However, Martin et al. [8] stated that the prescriptive drainage systems are not critical in achieving a speci ed factor of safety.
According to Ahmed et al. [2], the earliest usage of horizontal drains was recorded in 1843 in Great Britain to stabilize railway slopes which involved very deep cuts, while in the USA, its usage was rst reported in 1939, but the method only gained acceptance within the continent in the early 1970s.Stanton [9] reported the successful use of horizontal drains to numerous landslides by the California Division of Highways.In Australia, its usage was rst reported in the mid-1960s, while in France it was rst used in 1954.Craig and Gray [10] found that, in Hong Kong, horizontal drains have been used at shallow depths, rarely more than 20 m in length since 1973.However, it is observed that a number of slope drainage systems utilizing horizontal drains in Malaysia did not function properly in the long term and drained the water ine ciently from slopes.Clogging of drainage pipes by soil sediments which leads to reduced e ciency in draining water is a typical condition that inevitably could cause slope failures.Hence, there is a need to understand the in uence of these technical parameters to help improve the existing prescriptive design.In Malaysia, manuals available for designing the pipe drains are Urban Stormwater Management Manual for Malaysia (MSMA) and Irrigation and Agricultural Drainage Manual from the Government of Malaysia's Department of Irrigation and Drainage (DID).
is paper focuses on evaluating various critical parameters in the design of slope drainage system.e identi ed parameters, namely, the drainage pipe con guration, lter design, soil types, and pipe arrangement, were tested for their in uence towards the drainage system's e ciency.All experiments were performed in the Geotechnical Engineering Laboratory, Universiti Malaysia Sarawak (UNIMAS).Several conditions were maintained throughout the experiments: the soils (1) were kept homogeneous and (2) were of type cohesive and noncohesive.Rainfall simulator was used to simulate typical heavy rainfall condition in Malaysia.

Materials and Methods
e soils used in this study were sand and lateritic soil.A number of lter materials were utilized, namely, quarry gravels and oil palm shells, as aggregate-type lters, while geotextile and nylon square net fabric were used as envelopetype lters.
e pipe chosen was perforated high-density polyethylene (HDPE) pipe.All tests were performed in a model box equipped with a rainfall simulator.
Sand was sieved with 1.18 mm sieve, and the particle size distribution for the sand is as presented in Figure 1. e coe cient of uniformity, C u , was 2.035, which indicates that the soil is uniformly graded.For the lateritic soil, C u was 19.06, and the coe cient of curvature (C c ) was 2.26. Figure 1 presents the particle size distribution of the soil.According to Ishibashi and Hazarika [11], the lateritic soil was classi ed as silty soil with high plasticity by referring to the USCS chart.It was well graded as it satis ed the C u ≥ 6 and 1 ≤ C c ≤ 3 requirement.Based on Atterberg's limit test, the plasticity index and liquid limit of the lateritic soil were 45% and 95%, respectively.e targeted moisture content for the sand and lateritic soils to overcome the water repellence problem was 18.6% and 22%, respectively.Hence, the unit weights of the wet sand and lateritic soils were 14 kN/m 3 and 12 kN/m 3 , respectively.
e proposed lter materials were quarry gravels with size in the range of 1.18 mm to 1.70 mm and oil palm shells with size in the range of 1.70 mm-3.35mm.According to Terzaghi's lter criteria [12], the particle diameter of 15% size of the lter material (d 15 ) should be greater than 4 times the diameter of 15% size of the soil material (D 15 ) and smaller than 4 times the diameter of 85% size of the soil material (D 85 ).erefore, the required particle size for a lter material is between 0.77 mm and 1.88 mm to comply with the criteria.e gravel lter satis ed the criteria, whereas the oil palm shells did not due to limitation of crusher machine.However, a prospect of using waste materials as lter is seen interesting to be investigated and was maintained.Figure 2 shows the lter materials utilized in this study.
e HDPE pipe in the experiment was designed and scaled based on the Simona design [13].e Simona design incorporates partial perforation.However, some modi cations from the design were performed as to suit to the experiment requirement.e HDPE pipe was chosen as it has excellent resistance to corrosion, long service life, chemical resistance, and high abrasion resistance which make it ideal for water supply and drainage.e shape of perforation was circle with a diameter of 4 mm, and the distance between center and center of the holes was 70 mm as shown in Figure 3(a).e perforations in the pipe were located in the upper section of the pipe (partial drainage pipe) and were symmetrical to the pipe's vertical axis at an angle of 220 degrees (Figure 3(b)).e partial drainage pipe was proposed because of the nature of the study in which the 2 Advances in Civil Engineering pipe was used to absorb strata and surface water through the perforation at the top and convey it to the next receiving water course in the enclosed bottom of the pipe.Table 1 shows the details of the proposed partial drainage pipes used in this study.
Geotextiles with an opening size of 0.12 mm were used as lter and pipe envelopes to prevent the movement of ne particles into the pipe.In addition, another material with larger opening (1 mm), nylon square net fabric, was also proposed.
A medium-sized experiment model box was used for the modeling, and it was scaled based on the pipes installed in a slope at Nanyang Heights in Singapore [14], which in detail is presented in Table 2. Nanyang's pipe arrangement was followed considering its similar tropical setting as in the case study.e dimensions of the model box are 0.8 m long, 0.6 m wide, and 0.6 m high (Figure 4).A steel frame is used to support the whole box structure.
e bracing steel is to support the sides of the model box.e model box is designed to be leak proof.
A rainfall simulator shown in Figure 5 was to provide similar characteristics to the natural rainfall.A submersible pump (60 liter/minute (LPM) capacity) placed in a water tank was used to pump water to feed the rainfall simulator.Simulated rainfall was introduced through a water sprinkler system at a constant rate of water ow regulated by a ow meter installed in the system.Measurement of rainfall intensity was taken by using the spatial distribution method [15].A rectangular test plot of 0.48 m 2 (the size of actual plot) was located under the rainfall simulator and was introduced to rainfall for 5 minutes.Various falling heights (between nozzle and bottom of measuring cups) and ow rates were introduced before a nal height of 57 cm, ow rate of 18 LPM, and appropriate rainfall dispersion angle of 45 °29′33″ to cover the whole plot were reached.e simulated rainfall's intensity was attained by measuring the  height of water collected from the individual measuring cup as shown in Figure 5(b).Average rainfall intensity obtained for this study was 434.1 mm/h.Table 3 presents the mass of materials used in all seven experiments performed in this study.e mass of sand and water for mixing purposes in experiment with single pipe was 167.28 kg and 31.11kg, respectively.e mass of sand and water in experiment with double pipes was 161.76 kg and 30.09kg, respectively, and the mass of lateritic soil and water in experiment with single pipe was 139.41 kg and 30.67 kg, respectively.e varied masses were due to initial water content of the soil, and all materials were mixed with an aim to achieve the targeted wet density as mentioned in the earlier section.
Prior to securing the perforated HDPE pipe in the model box, a square caging of 80 × 80 mm was prepared to house the pipe.Aggregate lter was lled into the caging and was wrapped with a layer of geotextile.e perforated HDPE pipe was also wrapped with a 6 mm thick geotextile lter envelope, and the wrapped pipe was later placed in the caging which was already partially lled with aggregate lter.
e ready HDPE pipe enclosed in a caging lled with geotextile-wrapped aggregate lter is shown in Figure 6(a).
e pipe in its caging was then attached to the bottom outlet of the model box shown in Figure 6(b), and the soil mixture was then placed at a height of 300 mm.To attain the desired density, it was identi ed from a separate compaction test that applying rm hand pressure of about 25-30 presses for every soil lift of 50 mm thickness was su cient.Also, to provide the 5-degree inclination of the pipe, an inclinable board resting on the bottom of the box was placed at the rear of the pipe.Once the soil model was ready, the rainfall simulator was then placed at the top of the model box as shown in Figure 6(c).Figure 7 shows the schematic diagram of the experiment's con guration.During the experiment, the mass of water discharged from the pipe was recorded every 1 minute for the duration of 7 minutes.After this, the remaining mass of water discharged was recorded in total amount at the end of discharge.Sediments in the collected water were weighed for their dry mass, and permeability of the soil was measured   4 shows the di erent scenarios introduced in the 6 experiments performed in this study.
e di erences between these 6 experiments are the types of soil, envelope lter material, aggregate lter material, quantity of perforated pipes installed, and types of pipe envelope.

Results and Discussion
Experimental results are discussed based on comparable scenarios between experiments, and Table 5 shows the summary of important ndings which are the average discharge (Q; m 3 /s), percentage of water discharged, amount of sediments collected (g), and permeability of soil before/after experiments (cm/s).
Experiments 1 and 6 are compared based on soil type.Both soils were mixed with their respective amount of water to achieve the targeted density.ere was no sediment collected in the discharged water from both experiments.Peak discharge rates for Experiments 1 and 6 are 1.93 × 10 −3 m 3 /min and 2.32 × 10 −3 m 3 /min, respectively.Both peak rates were recorded at about 120-150 seconds.Experiment 1 (sand) shows higher discharge rate compared with the lateritic soil model in Experiment 6 in the rst 1.5 minutes.However, despite low permeability of the lateritic soil model, it can be observed after 1.5 minutes the discharge rate of the lateritic soil model is higher than in the sand model.In addition to that lateritic soils were able to drain more in ltrated water after sometime compared to sand.It was identi ed that this interesting condition appeared due to existence of retained or trapped water above the drainage pipe in the lateritic soil and on the surface of the soil.Its low permeability contributed to this interesting condition, and this can be observed in Figure 8.
Experiments 1 and 4 are compared with respect to the drainage envelope.
ere was sediment collected in the discharged water from drainage system enveloped with net (Experiment 4).It can be observed from Table 5 that the discharge rate of Experiment 4 is higher than in Experiment 1 (with geotextile).e e ective opening size of geotextile and net is 0.12 mm and 1 mm, respectively.From the sieve analysis of sand, the e ective opening size of the net allows the biggest sand particle and water to easily pass through and

6
Advances in Civil Engineering thus resulted in the sediment collected at the end of the experiment.e requirement of a lter is to keep soil particles from invading the lter signi cantly.Hence, the proposed net is not appropriate to be used as a lter envelope in a drainage design.
Experiments 1 and 2 are compared in terms of aggregate lter material.ere was sediment collected in the discharged water from Experiment 2 (oil palm shell).Peak discharge rates for Experiments 1 and 2 are 1.93 × 10 −3 m 3 /min and 2.36 × 10 −3 m 3 /min, respectively.Both peak rates were recorded at about 120 seconds.However, it can be observed that the average discharge rate of Experiment 1 (quarry gravel) is higher than that of Experiment 2. From the particle size analysis on both soils, the particle size at 15% of oil palm shells is higher than that of the quarry gravels.Hence, the permeability rate of the oil palm shells is expected to be higher than that of the quarry gravels as the particle size of the oil palm shells is larger than that of the quarry gravels.Consequently, the porosity of the oil palm shells is also higher than that of the quarry gravels.Hence, the sediments recorded in Experiment 2 may be due to some of the soil particles which could have passed through the oil palm shells and owed through the perforated pipe.In addition to that the particles in the quarry gravels and oil palm shells were not of the same size and shape as the colloidal theories assumed.Oil palm shells consist of bres which allow clogging to occur at the outer surface of the perforated pipe; thus, sand was retained at the bre of oil palm shell which led to much lower rate of water discharged.Figure 9 shows the rate of water discharge.
Experiments 1 and 3 (2 pipes) are compared in terms of quantity of pipes installed in the model box.e sediment collected in the discharged water from experiment with double perforated pipes was found to be higher than in the experiment with single pipe.Based on Table 5, the discharge rate of experiment with double perforated pipes was higher than in the experiment with single perforated pipe. is is logical since the number of pipes was increased, the discharge rate of water was also increased.Similar ndings were also shown in Experiment 5 in which higher discharge and sediment were found in double perforated pipes with oil palm shells as aggregate lter.Peak discharge rates for Experiments 1 and 3 are 1.93 × 10 −3 m 3 /min and 3.55 × 10 −3 m 3 /min, respectively.Both peak rates were recorded at about 60-120 seconds.Meanwhile, peak discharge rates for Experiments 2 and 4 are 2.36 × 10 −3 m 3 /min and 3.56 × 10 −3 m 3 /min, respectively.Both peak rates were also recorded at about 120 seconds.Figure 10 shows the comparison.
From the comparisons made above, it is recommended that the pipe installed in the slope could be wrapped with geotextile and in tandem with application of granular lter to minimize clogging without a ecting the discharge rate of water. is is evident from the ndings in Experiments 1 (geotextile) and 4 (net) in which lower amount of sediments and good amount of water discharged were recorded in Experiment 1 which had combination of geotextile and gravel lter in a system.
Terzaghi's lter criteria could be followed closely when deciding on new materials to act as aggregate lter and ndings in Experiments 1 (gravel lter) and 2 (OPS lter) support this statement.Some sediments were collected in Experiment 2 as OPSs had sizes not within the recommended range.Also, the brous nature of the material led to lower amount of discharge rate.
A caging system could be introduced as it could maintain the integrity of the drainage system and could ease installation.
e 5-degree inclination of pipe is highly recommended because it provides higher hydraulic head and lower piezometric surface elevations at the back of the drains.Table 6 shows the comparison and recommendations made based on identi ed parameters and their in uences.Comparison was made between the Malaysian DID manual, the European standard, and this study.

Conclusions and Recommendations
e aim of this study was to identify the parameters which are important in designing slope drainage system and to recommend a design suggestion incorporating identi ed parameters and their in uences.e following conclusions can be drawn from the study: (i) Based on the literature review and 6 experiments performed, the parameters which are important in designing slope drainage system are angle of pipes installed, spacing between pipes, size and arrangement of pipes, pipe perforation, and lter materials.(ii) e proposed lter materials which are oil palm shells and net did not perform well in terms of sediment collected as they do not comply with the relationship between particle size and e ective opening size. is criterion must be observed closely when deciding new lter materials.(iii) In terms of sediment collected in discharged water, the pipe was enveloped with geotextile, and quarry gravels in both soils were able to reduce clogging in the pipe.(iv) In terms of discharge rate and percentage of water discharged, the void ratio of the soils and granular lter material played an important role because the higher the void ratio, the higher the permeability and the discharge rate.However, in cases where permeability is low, amount of water discharged may be signi cant after sometime as found in this study.(v) By comparing with the Malaysian DID manual, it is recommended that the pipe to be installed in the slope is to be wrapped with geotextile to minimize clogging.e inclination of the pipe from the horizontal is recommended because it provides higher hydraulic head and lower piezometric surface elevations at the back of the drains.A study is recommended on nding a suitable range of angles.(vi) Long-term performance of the drainage system could be further investigated.Critical factors that can be identi ed from this study are (1) structural integrity of the drainage system which could be improved via the caging system and (2) clogging of the drainage system at the lter.

Figure 1 :Figure 2 :
Figure 1: Particle size distribution of sand and lateritic soil in this study.

Figure 3 :
Figure 3: Details of perforations in the HDPE pipe.(a) Placement of perforations on the HDPE pipe and (b) partial pipe drainage.

Figure 6 :Figure 7 :
Figure 6: (a) Pipe in caging.(b) Setup of pipe in the model box.(c) Experimental setup (not to scale).

Figure 8 :
Figure 8: Comparison of discharge between Experiments 1 and 6 for the rst 7 minutes.

Figure 9 :
Figure 9: Comparison of discharge between Experiments 1 and 2 for the rst 7 minutes.

Figure 10 :
Figure 10: Comparison of discharge between (a) Experiments 1 and 3 and (b) Experiments 2 and 5 for the rst 7 minutes.

Table 2 :
Scale factor between pipes in this study and Nanyang.

Table 3 :
Materials used for single-and double-pipe experiments.

Table 4 :
Details of experiments performed in this study.

Table 6 :
Recommendation between study, European standard, and DID manual.Geotextile with at least 5 mm thickness.Comply with ISO 9863, 9864, and 12956 in terms of mass per unit area, pore size index, and wettability