Design and Preliminary Results of Debris Flow Analog Model in Laboratory Scale, Case Study of Bentarsari Sub-watershed

Bentarsari sub-watershed is located at Brebes Regency administrative area, Central Java. This sub-watershed is surrounded by hills composed by volcanic breccia of Kumbang Formation (Tpk) and consists of sedimentary rocks at the lower part of the watershed. Volcanic breccia is the oldest rock in the watershed and appears to have a more intensive weathering process, making it more susceptible to landslides which are mostly followed by debris flow. Accordingly, the aim of this research is to conduct a laboratory scale analog modeling of debris flow event in this sub-watershed. The main materials used for modeling are moderate to completely weathered volcanic breccias to residual soils. The rainfall scenario is 50 mm/day as lower bound of heavy rain, 100 mm/day as lower bound of very heavy rain and 150 mm/day as lower bound of extreme rain. As a first step, the computer simulation is used for analyzing material movement of the model, using spherical materials with a density of 1.84 g/cm3 taken from average value of laboratory test results. The simplification process is carried out on spherical material with 4 cm and 2 cm diameter to depict moving material. The solid material volume used in the simulation is 0.07 m3 while the liquid is 0.03 m3. Then, a laboratory scale analog model has been made with a tub having dimensions of 0.5 m high, 2 m long and 1 m wide. When this analog model is tilted 50 degrees, the resulting slip plane then extends to the backward of the slope and the flowing materials move downward colliding with the previously deposited material, forming a flow which bends slightly to the left at the foot slope. The results of this research are expected to lead to a new method for modelling debris flows using analog models at laboratory scale.


Introduction
Indonesia is a country which is very prone to landslide disasters.Many landslides occur in the areas with steep to very steep slopes, intensive weathering processes, and high rainfall intensity.Landslide which has a mixture of solid and liquid materials can turn into a debris flow.Several landslides were followed by debris flow, including the Cililin landslide on April 21, 2004, the Jember landslide on January 2, 2006, the Ciwidey landslide on February 23, 2010, the Karangkobar landslide on December 12, 2014, and the Hululais landslide on April 28, 2016, have claimed many lives and caused extensive damage.For example, the debris flow event in Jemblung, Sampang Village, Karangkobar District, Banjarnegara Regency, Central Java, or what is often known as the Karangkobar Landslide, has claimed more than 100 lives, dozens of residents were injured, buried around 40 houses and cut village roads reaching 500 m [1].Another example is Pasir Panjang debris flow event at Bentarsari sub-watershed, Salem District, Brebes Regency, Central Java on February 22, 2018 which has claimed 18 lives, cut off the road connecting the districts and damaged hectares of rice fields.The aim of this research is to conduct a laboratory scale analog modeling of debris flow event in this sub-watershed, especially at the northern slope of where Pasir Panjang debris flow occurred.

Debris Flow
According to Cruden and Varnes (1978) [2] landslide can be divided into six categories based on the type of movement, which are falls, topples, slides, lateral spread, flows, and combined (complex) type.According to Stiny (1910) [3], the debris flow begins with flooding on steep mountain slopes and transports large amounts of suspended load and bed load.When the amount of sediment carried by the flow increases to a certain extent, the flows change into a viscous mass of water, soil, sand, gravel, stones and logs mixed together through the valley channel.According to Varnes (1978) [4] debris flow is a mass containing solid material, water, and air flowing as a liquid stream.The debris flow has three parts, namely the source area, the flow track and the run-out lobe/depositional area.
The source area is a place for accumulation of debris flows initial material which found on the upper slopes.The accumulation of debris flow initial material can be in the form of slope-forming material or forming natural dams.Natural dams are formed when an initial landslide occurs in a source area which blocks an existing river.The blockage which occurs then produces a puddle that saturates the clogging materials.The saturated materials will reach a critical point for a landslide to occur followed by debris flow.Apart from the presence of natural dams, the debris flow can also occur through sudden loading and liquefaction.The occurrence of landslides above the river slopes causes a sudden loading on saturated river deposits.This condition will initiate liquefaction in river deposits.The flowing liquefaction process developed into a debris flow with material originating from landslide and river sediments (Figure 1).The flow track is the place where the debris flow flows toward the valley associated with the river flow.The deposition area is a plain as a place for the deposition of debris flow materials.At the boundary between the flow track and the deposition area, the debris flow will form a fan formation.The distribution of debris flows in the deposition area depends on the topography of the deposition area.If in the depositional area still has terrace geometry, then a second flow path can be formed which will form a second depositional area (multiple depositional area).Debris flow often occurs in areas which have hilly morphology with high rainfall.The slope angle and the river channel are determining factor in the process of debris flow velocity.Debris flow will occur if there is an abundant supply of debris material with a large supply of water as flowing medium.Debris flow most frequently occurs after or during heavy rains.The debris flow has high specific gravity so that large chunks can be carried by the flow to the end of deposition area.
Debris flow generally follows pre-existing surface flows but can move down slopes and across unobstructed alluvial fan surfaces.This is because the debris flow tends to build its own channel due to the natural embankments which form at the lateral flow boundaries.The debris flow moves down the slope in a series of waves with periods ranging from a few seconds to several hours.The debris flow front is higher than the other sections and contains large moving boulders.The frontier flow is followed by more fluid and turbulent mud with high concentrations of suspended sediment.This more liquid phase continues until the next wave arrives or until the debris flow activity ceases [6].The speed of the debris flow varies because it is influenced by the size of debris material, concentration, sorting of debris material, and channel geometry which includes shape, slope, width, and curve.
The debris flow can be highly erosive as it moves in the flow track [7].During movement in the flow track, the density of the debris flow can be twice as large as the density of the accompanying fluid.The debris flow can also increase the shear stress up to 6 times at the bottom of the river flow.There is an increase in shear stress for 20 seconds after the front of the debris material enters the flow track [8].The increase in shear stress is caused by a sudden loading at the location of landslide in the source area.An additional attractive characteristic of debris flows is the ability to move boulder over great distances.The mobility of the debris flow is highly dependent on the presence of clay-sized material (even only 1% -2%).Clay-sized materials reduce permeability and increase pore pressures, thereby increasing the mobility of the larger materials.

Analog Model
The analog model has been made with dimensions of 1.75 m height, 2 m length, and 1 m width (Figure 3).A tub with dimensions of 0.5 m high, 2 m long and 1 m wide and with one side open is placed on a support.This tub will be used to place the debris materials for modeling analogously on a laboratory scale.The tub will be supported by an electric hydraulic pump capable of tilting the table up to 75º (Figure 4).The hydraulic pump is expected to be able to tilt the tub which has been filled with 2 tons of debris flow materials.The Bentangsari sub-watershed is surrounded by hills composed by volcanic breccia of Kumbang Formation (Tpk) and consists of sedimentary rocks at the lower part of the watershed.Volcanic breccia is the oldest rock in the watershed and appears to have a more intensive weathering process, making it more susceptible to landslides which are mostly followed by debris flow.The slope to be modeled in this research is the northern slope of the Bentangsari sub-watershed which is composed of the Kumbang Formation (Tpk).This formation is then in detailed classified into five rock and soil units, which are Fresh -Slightly Weathered Volcanic Breccia Unit, Fresh -Slightly Weathered Sandstone Unit, Moderate -Highly Weathered Volcanic Breccia Unit, Medium -Highly Weathered Sandstone Unit, and Completely Weathered -Residual Soil of Volcanic Breccia unit.The main materials for modeling debris flow are Moderately -Strong Weathered Volcanic Breccia Unit and Completely Weathered Volcanic Breccias -Residual Soil Unit.
Hydrograph supply (water) will affect the flow viscosity so that the rheological conditions of the research area can be ascertained.The rheological property of the debris flow is the relationship between deformation and flow due to shear stress changes.Rheology can be approximated by the parameter of viscosity (concentration).The rheology of the flow will describe the general bulk density, flow strength, fluid type, and the resulting landscape and deposits.The nearest rainfall station to the research area is PK.5c Dukuhjeruk (± 25 km).The PK.5c Dukuhjeruk monthly rainfall value is used as the basis for determining the rainfall scenario which will be used in the analog model.PK.5c Dukuhjeruk has a maximum rainfall of 493 mm/month which occurred in March 2010, has an average rainfall range of 24 mm/month -369 mm/month in 2009-2015, and there are no rain events in July -September (Table 1 and Figure 5).If it is assumed that the rain that occurs in one month accumulates on certain days, the rainfall scenario which will be used is 50 mm/day as lower limit for heavy rain class, 100 mm/day as lower limit for very heavy rain class and 150 mm /day as lower limit of extreme rain.

Preliminary Result
The first analog modeling experiment used materials taken from the Pasir Panjang debris flow.As a first step, the computer simulation is used for analyzing material movement of the model, using spherical materials with a density of 1.84 g/cm 3 taken from average value of laboratory test results.The simplification process is carried out on spherical material with 4 cm and 2 cm diameter to depict moving material.A laboratory scale analog model was then carried out using material with a volume of 0.07 m 3 and supply hydrograph with 0.03 m 3 water volume which added slowly (Figure 6).The comparison between the solid and liquid volumes of 7:3 is based on previous research result [9].The tub slope angle slowly increases until 20º, a failure occurs which causes the saturated materials to move downwards.This movement only occurs in saturated materials, so the slip planes which occur are on the boundary of the saturated and unsaturated zones (Figure 7).When the tub is tilted 50°, the slip plane then extends

Concluding Remarks
A laboratory scale analog model using a tub with dimensions of 0.5 m high, 2 m long and 1 m wide, with a volume of 0.07 m 3 material and supply hydrograph with 0.03 m 3 water volume shows that when the tub with 20º slope angle, a failure occurs which causes the saturated material to move downward with slip planes on the boundary of saturated and unsaturated zones.When the tub is tilted 50°, the slip plane then extends to the backward of the slope and the flowing materials move downward colliding with the previously deposited material, forming a flow which bends slightly to the left at the foot slope.The results of this research are expected to lead to a new method for modelling debris flows using analog models at laboratory scale.For further improve the reliability of the analog model results, subsequent modeling is further suggested to be carried out with the same material by adding several parameters, such as artificial rain tools, artificial slope surface profiles, and static cameras at four locations to record the movement of the modeled materials.

Figure 1 .
Figure 1.Illustration of the initiation of debris flow due to sudden loading and liquefaction [3].

Figure 3 .
Figure 3. Debris flow laboratory scale analog model design.

Figure 4 .
Figure 4. Electric hydraulic pump which is under the tempered glass.
of the slope and the flowing materials move downward colliding with the previously deposited material, forming a flow which bends slightly to the left at the foot slope (Figure8).

Figure 6 .
Figure 6.The initial conditions of the analog model.
[1] Sadisun I A Kartiko R D and Dinata I A 2018 Numerical Simulation of Some Debris Flow Events in Central Java for Predicting Run-out Distributions Proceedings of SEAGS-AGSSEA Conference 2018 (Jakarta) pp 357-360 [2] Cornforth D H 2005 Landslides in Practice: Investigation, Analysis, Remedial/Preventive Options in Soil, (Hoboken: John Wiley & Sons) [3] Jakob M and Hungr O 2005 Debris-flow Hazard and Related Phenomena (Chichester: Praxis Publishing Ltd) [4] Blijenberg H M 2007 Application of physical modelling of debris flow triggering to field conditions: limitations posed by boundary conditions Engineering Geology 91 pp 25-33 [5] Nettleton I M Martin S Hencher S and Moore R 2005 Debris flow types and mechanisms (Scottish Road Network Landslides Study) ed Winter M G MacGregor F and Shackman L (Edinburgh) pp 45-67 [6] Pierson T C 1980 Erosion and deposition by debris flows at Mount Thomas, North Canterbury, New Zealand Earth Surface Process 5 pp 227-247 [7] Janda R J Scott K M Nolan K M and Martinson H A 1981 Lahar movement, effects, and deposits (The 1980 eruptions of Mount St. Helens) ed Lipman P W and Mullineaux D R (Washington) pp 461-478 [8] Berger C Schlunnegger F and McArdell B W 2011 Direct measurement of channel erosion by debris flows, Illgraben, Switzerland Journal of Geophysical Research 116 pp 1-18 [9] Sadisun I A Dinata I A and Kartiko R D 2019 Run-out Distribution of Pasir Panjang Landslides Followed by the Debris Flow, Central Java, Indonesia (Jeju: IAEG ARC12 2019)