sw 3524 ms
8/22/01
2:51 PM
Page 641
RAINFALL SIMULATOR FOR LABORATORY STUDIES
T. P. Regmi, A. L. Thompson
ABSTRACT. Numerous types of rainfall simulators have been used to study the fundamental behavior of sediment transfer
and chemical movement in soil from rainfall. For results of these studies to be representative, it is essential that the
artificial rainfall have characteristics similar to that of natural rainfall. In this study, a number of design modifications
were made to a stationary rainfall device in order to achieve rainfall simulation comparable to natural rainfall. A 1 m ×
1 m laboratory-scale rainfall simulator was designed and constructed using the positive volume displacement principle to
provide a precise and broad range of rainfall intensities from 0.025 to 16 cm/h in increments of 0.025 cm/h. The dropforming mechanism consisted of a telescopic arrangement of 21 and 9 gauge stainless steel tubings, producing a mean
droplet diameter of 4.5 mm. A drop-size distribution with characteristics similar to natural rainfall was produced when
droplets passed through a suspended drop redistribution screen. When the droplet size spectrum for different height of
screen suspension was mapped with the spectrum produced by different intensities of natural rainfall, it was possible to
estimate a height of suspension corresponding to a desired application rate with characteristics similar to natural rainfall
at that given rate. The fully automated rainfall simulator was arranged in a closed loop using a feed tank and automated
solenoid valves. The entire rainfall simulation system was installed at a height of 14 m to allow droplets up to 4.3 mm
diameter to reach 95% of their respective terminal velocities. Flow rate variation among the 216 individual drippers,
used in the rainfall simulation design, was less than 15% of the mean at system application rates less than 1.5 cm/h, and
less than 9% at rates greater than 8 cm/h. Reproducibility of system application rate for a given air inlet tube position
was within 1%. The best commercial unit available had an advertised ± 20% variation in system application rate with
only 50% of the range in rate of the unit developed here.
Keywords. Rainfall intensity, Rainfall simulator, Telescopic drop formers, Terminal velocity, Volume displacement.
R
aindrop energy is known to have a major
influence on soil erosion (Wischmeier and Smith,
1978). The transfer of chemicals to runoff from
the subsurface is thought to be a result of
turbulence in soil water generated by the impact of
raindrops (Ahuja et al., 1983; Ahuja, 1990). Using physical
principles, relationships have been quantified to predict the
interactions among application rate, droplet size, kinetic
energy, and droplet energy flux (Thompson and James,
1985). However, the relationship between chemical transfer
and movement into the soil, and the interaction with
infiltration through a developing surface seal, is not well
understood (Ahuja and Lehman, 1983).
Numerous types of rainfall simulators have been used to
study the fundamental behavior of sediment transfer and
chemical movement in soil resulting from rainfall. The use
of rainfall simulators generally provides a more rapid,
efficient, controlled, and adaptable tool than natural rainfall
(Meyer, 1965). The influence of water droplet kinetic
energy can be examined by either varying fall height of the
Article was submitted for publication in January 2000; reviewed and
approved for publication by the Soil & Water Division of ASAE in July
2000.
The authors are Tulsi P. Regmi, Graduate Research Assistant, and
Allen L. Thompson, ASAE Member Engineer, Associate Professor,
Agricultural and Biological Engineering Department, University of
Missouri, Columbia, Missouri. Corresponding author: Allen L.
Thompson, University of Missouri, 251 Agricultural Engineering Bldg.,
Columbia, MO 65211, phone: 573.882.4004, fax: 573.882.1115, e-mail:
<thompsona@missouri.edu>.
droplets, changing droplet sizes, or protecting the soil
surface with a residue cover to absorb droplet energy.
Droplets from gravity-fed simulators all start at rest and
accelerate to a unique terminal velocity based on diameter.
The smaller the droplet, the lower the terminal velocity and
the less fall height required to reach that velocity (Laws
and Parsons, 1943). Droplet kinetic energy, which is
directly related to droplet size and the square of droplet
velocity at impact, will vary accordingly. The effect of
droplet kinetic energy on soil movement and the chemical
transfer process is critical, therefore both the size
distribution and fall velocity of droplets should be
simulated as closely as possible to natural rainfall.
Design of a rainfall simulator should address the control
of application rates in both time and space, the
reproduction of drop size distributions observed with
different intensities of natural rainfall at the corresponding
application rates, and the reproduction of the terminal
velocities of drops in natural rainfall (Hall, 1970; Mech,
1965). Application rate for a given drop former is
controlled by two methods: areal density of drop formers,
and hydrostatic head (Mutchler and Moldenhauer, 1963;
Mutchler and Hermsmeier, 1965).
Generally, rainfall simulators can be categorized into
two types: tubing-tip simulators and nozzl;e simulators
(Hall, 1970). If a nozzle-type rainfall simulator is used, an
increase in working pressure increases the average
intensity but decreases the range and mean of drop sizes
within the spray—just the opposite that is observed with
natural rainfall. Alternatively, if a tubing-tip rainfall
Applied Engineering in Agriculture
VOL. 16(6): 641-647
© 2000 American Society of Agricultural Engineers 0883-8542 / 00 / 1606-641
641
sw 3524 ms
8/22/01
2:51 PM
Page 642
simulator of fixed tube diameter is used, the result is the
production of uniform drop sizes (Hall, 1970). Ekern and
Muckenhirn (1947) used hypodermic needles enclosed
with various diameters of glass tubing to produce drops
ranging from 2.75 to 5.8 mm in diameter. Mutchler and
Moldenhaur (1963) used drop formers consisting of 25 mm
lengths of four different gauges of hypodermic tubing
telescoped with overlap (9, 10, 12, and 14 gauge). The
smallest tube controlled the flow and the largest tube
produced drops varying from 5.02 to 5.45 mm in diameter
depending upon the head. Bryan and De Ploey (1983) used
capillary-tube drop formers that produced 4.1 to 4.2 mm
drops. The drop former bed was installed at a height of
7.1 m with a 3.07-mm square mesh drop redistributionscreen suspended 1.7 m below the drippers. This drop
redistribution-screen reworked the 4.1 to 4.2 mm drops to
give the following drop size distribution: 0 to 1 mm
(2.2%); 1 to 2 mm (29.4%); 2 to 3 mm (41.4%); 3 to 4 mm
(23.2%); 4 to 5 mm (3.8%). The mean diameter of the
redistributioned drops was 2.3 to 2.4 mm, which was
somewhat smaller than natural rainfall.
Generally, the major difficulties encountered in
designing a rainfall simulator are: (1) determination of an
appropriate density of drop formers; under conditions of
constant head, intensity of application increases as the
spacing of drop formers decreases; (2) uniformity of
application; (3) droplet size distribution; (4) consistent
control of application rate; and (5) reaching terminal
velocity for the droplet distribution. Therefore, the
objective of this study was to design and construct a
laboratory-scale, stationary rainfall simulator that would
produce droplet characteristics similar to that of natural
rainfall.
SYSTEM DESIGN
Three fundamental criteria are commonly considered in
designing a rainfall simulator (Hall, 1970), namely:
• The control of application rates in both time and
space.
• The reproduction of drop-size distributions observed
in different intensities of natural rainfall at the
corresponding application rates.
• The reproduction of the terminal velocities of drops
in natural rainfall.
A laboratory-scale rainfall simulator was constructed as
shown in figure 1. The design considerations included the
use of telescopic drop formers and drop redistribution
screen, and the hydraulic principle of positive volume
displacement. The telescopic drop formers used in the
design consisted of differential lengths of two gauges of
stainless steel hypodermic tubing to control the drop size
and rate of application (fig. 2). The hypodermic tubing
used for the simulator was 5 cm lengths of 21-gauge
capillary tubing (for flow control) and 2 cm lengths of
9-gauge tubing (drop size control) placed in a telescopic
arrangement in 1.9-cm long (2.4-cm long including bolt
head), 0.79-cm-diameter bronze bolts as illustrated in
Figure 1–Design schematic of the laboratory-scale rainfall simulation system.
642
APPLIED ENGINEERING IN AGRICULTURE
sw 3524 ms
8/22/01
2:51 PM
Page 643
Figure 2–Design schematic of the telescopic drop former.
figure 2. Metal burrs were removed from the ends of all
tubing after cutting to desired lengths. The head end of
each brass bolt was drilled 1.5 cm deep for housing the
outside diameter (OD) of the 9-gauge tubing, with the
remaining bolt length drilled to house the 21-gauge tubing.
The 21-gauge tubing was inserted into the bolt, even with
the bolt head and overlapping 1.5 cm of the 2-cm-long
9-gauge tubing. Both the 21- and 9-gauge tubes were
secured to the bolt with epoxy as shown in figure 2.
Operational design of this laboratory-scale rainfall
simulator is based on the principle of positive volume
displacement in an otherwise sealed container. This
principle simply states that ∆Va = ∆Vf at constant pressure
head, where ∆Va = change in volume of air entering the
system and ∆Vf = change in volume of liquid leaving the
system. The pressure head in the tank is measured from the
bottom of the air inlet tube to the lower tip of the drop
former. The point of air entry at the bottom of the air inlet
tube registers atmospheric pressure (P atm ) within the
dripper tank. Hence, the water head above the drop formers
is manipulated by raising and/or lowering the air inlet tube.
For example, when the tube is raised, the distance from the
outer tip of the drop former to the bottom of the tube
increases, increasing the pressure head above the drop
formers, and vice versa. Because flow rate varies
proportionately with head, this provides a precise control
of application rates (fig. 1).
In a pre-design pilot study, a relationship between flow
rate and static head for different diameters of telescopic
drop formers was quantified as described by figure 3. The
results indicated that flow rate varied greatly with the size
of the capillary tubing used, and was described by the
power function (Q = kP x), where k and x are coefficients
and Q and P are flow rate and static pressure, respectively.
VOL. 16(6): 641-647
The larger the capillary-tube size, the steeper and more
non-linear the slope of the curve. The drop formers with
22-gauge tubing showed very little change in flow rate
with head. For capillary tubes 21-gauge and larger, flow
resistance was considerably lower providing greater
flexibility in applying a wide range of intensities for a
small change in head. The 9-gauge tubing was selected for
the outlet tip because it produced a maximum droplet
diameter of 5 mm similar to Mutchler and Moldenhaur
(1963).
Figure 4 describes the relationship between the ratio of
flow rate to head as a function of length of the 21-gauge
capillary tube selected for flow control. This ratio is unique
for each tube diameter and water head as a function of tube
length. Figure 4 can be used to determine the resulting
system application rate for any dripper spacing by dividing
the flow rate per dripper by the area of the desired dripper
spacing. As this ratio increases, flow control decreases.
Tube lengths less than 2.5 cm provided too limited flow
rate control. Tube lengths greater than 7.5 cm had a much
smaller ratio, but this resulted in a very restricted range of
intensity rates (0.2 to 8.5 cm/h at static heads up to 50 cm
with drop former spacing of 7.7 cm). Rainfall application
rates commonly used in erosion and chemical transfer
studies are between 3.2 and 13.4 cm/h (Ahuja, 1990).
Capillary tubing with 5 cm length provided an adequate
flow rate range for the selected equilateral dripper spacing
of 7.7 cm used in the design.
The dripper unit consisted of a stainless steel tank, 1m ×
1 m × 0.6 m, with 0.64-cm wall thickness which was
enclosed using a 1 m × 1 m, 0.95-cm-thick stainless steel
dripper bed module. All structural components of the
dripper tank were built with stainless steel because of its
excellent strength and anti-corrosion properties. The
bottom of all four sides was bent into a 5 cm lip at 90°
angle, and the top and all four sides of the tank were MIGwelded along the seams. The bed module was attached to
the lip using 0.79-cm-diameter bolts spaced 5 cm apart and
sealed with a rubber gasket. A total of 216 drippers were
screwed into pre-threaded holes of the bed module in a
geometric configuration of equilateral triangles, 7.7 cm
apart to minimize the number of drippers while
maintaining adequate range of intensity rates, raindrop
coverage, and application uniformity.
The tank was equipped with a graduated 20 mm inside
diameter air inlet tube (2-mm wall thickness) of clear
polyurethane material used to control application rates.
Polyurethane was chosen for its strength, smoothness, and
resistance to abrasion. The raising and lowering of the flow
control mechanism was accomplished using a fully
automated stepper motor coupled with a nylon rack and
brass pinion system with precision of 0.9 mm head
differential. Air seal of the flow control tube mechanism at
the tank entrance was ensured with a set of Teflon V-rings
set within a 1.9 cm thick stainless steel housing with
adjustable tension.
A stainless steel drop redistribution screen composed of
0.4 mm wires with 3.07-mm mesh (opening) size, similar
to Bryan and De Ploy (1983), was suspended below the
dripper tank to facilitate reworking drops into a broader
drop-size distribution spectrum. During this study, exact
placement of the screen was selected to produce drop-size
643
sw 3524 ms
8/22/01
2:51 PM
Page 644
Figure 3–Flow rate as a function of hydraulic head for different capillary tube diameters.
Figure 4–Ratio of flow rate to head as a function of capillary length for 21-gauge tubing.
644
APPLIED ENGINEERING IN AGRICULTURE
sw 3524 ms
8/22/01
2:51 PM
Page 645
distributions at different application rates which would
yield characteristics similar to natural rainfall.
The system was installed in the rain tower at a height of
14 m which was determined to be adequate to ensure 95%
of terminal velocity for droplets 4.3 mm in diameter and
smaller. The dripper tank was suspended with heavy-duty
steel chains (9.5-mm link diameter) from all four corners,
which were fastened to a custom-built steel saddle placed
on an existing I-beam at the top of the rain tower. An
electric winch, rated for a half-ton weight, was installed on
the supporting I-beam permitting lowering and raising of
the dripper unit for routine maintenance. The dripper tank
has sufficient water-holding capacity to apply 0.65 m depth
over a 1-m2 bed area. For tests requiring greater application
depths, water is supplied from a 1.89-m3 auxiliary tank
connected to the dripper tank through air and water
delivery lines, allowing continuous operation (fig. 1). The
rainfall simulation system is supplied with reverse osmosis
(RO) treated water.
The modified flour method of Kohl (1974) was
employed for measuring droplet size. Circular pans 21 cm
in diameter and 2 cm deep were filled with flour and
leveled with a straight edge. The flour pans were exposed
to the simulated rain for one second, and air dried for 24 h.
After drying, the flour pellets were separated using a set of
sieves ranging from 0.5 mm to 5 mm in mesh size and
weighed. Diameter of water droplets was computed using
the following relationship: water droplet diameter = 1.1 ×
sieve mesh size.
RESULTS AND DISCUSSION
At the beginning of each operation, the entire rainfall
simulation system is maintained in the absolute closed
mode, i.e., all solenoid valves are closed (fig. 1). The air
inlet tube is positioned at the desired height based on the
calibration curve for application rates (fig. 5). To begin
water application, the air inlet valve is opened and air
enters the tank as droplets begin to form. Each water
droplet is replaced by an equal volume of air entering the
system through the inlet tube. As air rises to the top of the
dripper tank, it is immediately transported into the feed
tank through the air conduit pushing the net volume of
liquid into the dripper tank, ensuring a constant application
rate throughout the application period (fig. 1). Rainfall can
be stopped by simply closing the air inlet valve.
The areal uniformity of application for individual
drippers describes how uniformly drop formers perform in
the spatial framework, and is quantified as the standard
deviation about the mean for the observations made in the
study. Results obtained during the flow simulation
indicated that the aerial flow uniformity improved with
increasing intensity rates. At an intensity rate as low as
1.25 cm / h , the areal flow uniformity had a standard
deviation of 15%; whereas, it improved to 7.5% at
8.5 cm/h. The non-linearity of the flow performance of
drop formers may be associated with overcoming the
internal friction forces and capillary surface tension.
Figure 5 shows the flow calibration curve for the rainfall
simulation system for the relationship between static head
and application rate. The system is capable of controlling
Figure 5–Application rate calibration curve of the rainfall simulation system.
VOL. 16(6): 641-647
645
sw 3524 ms
8/22/01
2:51 PM
Page 646
static head over the dripper bed in 0.9 mm increments, with
resulting intensity rates that range from 0.025 cm/h to
16 cm/h in increments of 0.025 cm/h, adequate for both
chemical transfer and erosion studies. System application
rates were reproducible to 1%.
Figure 6 shows the drop-size distributions obtained with
different distances of screen suspension below the droplet
formers as a function of intensity rate. The results obtained
from the flour method for the droplet size distribution
indicated that the drop formers produced an average drop
size of 4.5 mm. As shown by the experimental results with
no drop redistribution screen (fig. 6), about 70% of the
droplets were between 4 and 5 mm in diameter, with 90%
of all drops between 3 and 5 mm diameter independent of
the application rate. When the droplet size spectrum for
different height of screen suspension was mapped with the
spectrum produced by different intensities of natural
rainfall (Laws and Parsons, 1943), it was possible to
estimate a height of suspension for the desired intensity
rate, which would approximate the characteristics of
natural rainfall at that given rate. For example, the
suspension distance of 0.5 m, operated at 5.1 cm / h
(2 in./h), produced drop size characteristics very similar to
that of a 5.1-cm/h (2-in./h) natural rainfall (Laws and
Parsons, 1943). At the given height of suspension, the drop
redistribution screen reworked average 4.5 mm drops to
give the following drop size spectrum: 0 to1 mm (7.5%); 1
to 2 mm (24.5%); 2 to 3 mm (40.5%); 3 to 4 mm (24%); 4
to 5 mm (3.5%) with mean diameter of reworked drops
between 2.5 and 2.75 mm. Bryan and De Ploy (1983)
reported similar characteristics of simulated rainfall with
3.07-mm mesh size drop redistribution screen but at a
suspension distance of 1.7 m below the dripper bed.
Overall, the screen suspension distance of 0.3, 0.5, and
1.0 m closely simulated the droplet size distribution
spectrum corresponding to intensity rates of 15.24 cm/h
(4 in./h), 5.1 cm/h (2 in./h), and 1.27 cm/h (0.5 in./h) of
natural rainfall, respectively. The greater the screen
suspension distance, the finer the droplets due to greater
droplet impact force on the screen causing increased
droplet shear. Results also indicated that for a fixed screen
suspension distance, droplet size distribution of the
simulated rain was essentially constant for various
application rates, indicating the droplet size distribution
was solely a function of screen suspension distance.
In this study, the effect of oscillation of the drop
redistribution screen on the droplet size distribution pattern
was also evaluated. It was determined that compared to the
droplet rework of a stationary redistribution screen, the
screen oscillation operated at less than 2 strokes/s with an
approximate 5-cm stroke length did not substantially
enhance the droplet spectrum of the simulated rain. No
splash was observed near the sides of the drop
redistribution screen because the screen was slightly larger
than the effective dripping area of the drop former bed
module and droplet impact on the screen was not great
enough to produce excessive splash. The effects of droplet
drift during the descent were minimal because the rain
Figure 6–Droplet size volume distribution as affected by the screen suspension distance.
646
APPLIED ENGINEERING IN AGRICULTURE
sw 3524 ms
8/22/01
2:51 PM
Page 647
tower is completely enclosed limiting lateral air circulation
across the droplet trajectory.
SUMMARY
The objective of this study was to design and construct a
stationary laboratory-scale rainfall simulator to produce
droplet conditions similar to natural rainfall. The results
show that a system designed with the volume displacement
principle provides precise control of variable application
rates, and the use of telescopic drop formers combined with
suspended drop redistribution screen greatly enhances the
rainfall simulation and produces rainfall characteristics
similar to natural rainfall. The results of this study are
summarized as follows:
1. The rainfall simulation system was fully automated
with a precise control of within 1% for application
rates ranging from 0.025 to 16 cm/h.
2. The precision of the variable application rate was
0.9 mm of head, which translates to 0.025 cm/h.
3. Drop formers consisting of telescopic tubing
arrangement (9 and 21 gauge) produced an average
drop size of 4.5 mm. The redistribution screen
reworked drops to produce a drop-size distribution
similar to natural rainfall with intensity rates ranging
from 1.27 to 15.24 cm/h.
4. Droplet size distribution of the simulated rain did
not vary significantly with application rate for a
fixed suspension distance of the drop redistribution
screen.
5. Droplet size spectrum characteristics for the
simulated rain as a function of application rates was
approximated by varying the height of the screen
suspension.
6. The oscillation of the suspended screen did not
enhance the droplet size spectrum as compared to
the stationary suspension.
7. The entire rainfall simulation system was installed at
a height of 14 m permitting rain droplets of 4.3 mm
diameter and smaller to reach 95% of their
respective terminal velocities.
REFERENCES
Ahuja, L. R. 1990. Modeling soluble chemical transfer to runoff
with rainfall impact as a diffusion process. Soil Sci. Soc. Am. J.
54: 312-321.
Ahuja, L. R., and O. R. Lehman. 1983. The extent and nature of
rainfall-soil interaction in the release of soluble chemicals to
runoff. J. Environ. Qual. 12: 34-40.
Ahuja, L. R., O. R. Lehman, and A. N. Sharpley. 1983. Bromide
and phosphate in runoff water from shaped and cloddy soil
surfaces. Soil Sci. Soc. Am. J. 47: 746-748.
Bryan, R. B., and J. De Ploey. 1983. Comparability of soil erosion
measurements with different laboratory rainfall simulators. In
Rainfall Simulation, Runoff and Soil Erosion, CATENA
Supplement 4, Braunschweig, ed. J. De Ploey, 33-56.New
York, N.Y.: Elsevier Science Pub.
Ekern P. C., and R. J. Muckenhirn. 1947. Water drop impact as a
force in transporting sand. Soil Sci. Soc. Am. Proc. 12: 441-444.
Hall, M. J. 1970. A critique of methods of simulating rainfall.
Water Resour. Res. 6(4): 1104-1113.
Kohl, R. A. 1974. Drop size distribution from medium-sized
agricultural sprinklers. Transactions of the ASAE 17(5): 690-693.
Laws, J. O., and D. A. Parsons. 1943. Relation of raindrop-size to
intensity. Trans. Am. Geophys. Union 24: 452-460.
Mech, S. J. 1965. Limitations of simulated rainfall as a research
tool. Transactions of the ASAE 8(74): 63, 75.
Meyer, L. D. 1965. Simulation of rainfall for soil erosion research.
Transactions of the ASAE 8(63): 63-65.
Mutchler, C. K., and L. F. Hermsmeier. 1965. A review of rainfall
simulators. Transactions of the ASAE 8(67): 67-68.
Mutchler, C. K., and W. C. Moldenhauer. 1963. Applicator for
laboratory rainfall simulator. Transactions of the ASAE 6(3):
220-222.
Thompson, A. L., and L. G. James. 1985. Water droplet impact and
its effect on infiltration. Transactions of the ASAE 5(28): 15061510, 1520.
Wischmeier, W. H., and D. D. Smith. 1978. Predicting rainfall
erosion losses—A guide to conservation planning, USDA
Handbook No. 537. Washington, D.C.: USDA.
ACKNOWLEDGMENT. Acknowledgment is extended to the
University of Missouri Research Board, the University of
Missouri Agricultural Experiment Station, and to the
CSWQRU, USDA-ARS, Columbia, Missouri, for financial
and technical support of this project.
VOL. 16(6): 641-647
647
sw 3524 ms
8/22/01
2:51 PM
Page 648