MIT Open Access Articles

Development of a System Model for Low-Cost, Solar-

Powered Drip Irrigation Systems in the MENA Region

The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.

Citation: Sokol, Julia, Grant, Fiona, Sheline, Carolyn and Winter, Amos. 2018. "Development of
a System Model for Low-Cost, Solar-Powered Drip Irrigation Systems in the MENA Region."
Volume 2B: 44th Design Automation Conference.

As Published: 10.1115/detc2018-86297

Publisher: ASME International

Persistent URL: https://hdl.handle.net/1721.1/137996

Version: Final published version: final published article, as it appeared in a journal, conference
proceedings, or other formally published context

Terms of Use: Article is made available in accordance with the publisher's policy and may be
subject to US copyright law. Please refer to the publisher's site for terms of use.
 Proceedings of the ASME 2018 International Design Engineering
Technical Conferences and Computers and Information in Engineering Conference
IDETC/CIE 2018
August 26-29, 2018, Quebec City, Quebec, Canada

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

DETC2018-86297

DEVELOPMENT OF A SYSTEM MODEL FOR LOW-COST, SOLAR-POWERED DRIP

IRRIGATION SYSTEMS IN THE MENA REGION

Julia Sokol Fiona Grant

Global Engineering and Research Lab Global Engineering and Research Lab
Massachusetts Institute of Technology Massachusetts Institute of Technology
Cambridge, MA, USA Cambridge, MA, USA
Email: [email protected] Email: [email protected]

Carolyn Sheline Amos Winter, V.

Global Engineering and Research Lab Global Engineering and Research Lab
Massachusetts Institute of Technology Massachusetts Institute of Technology
Cambridge, MA, USA Cambridge, MA, USA
Email: [email protected] Email: [email protected]

ABSTRACT than commercially available software, which tends to have

Drip irrigation has the potential to conserve water and broader applications and focuses on larger systems. Future
increase crop yields. However, existing drip irrigation systems model generations will add more variables to the optimization
often require high pumping power, making them financially scheme—including pump selection, variable emitter flow rates
inaccessible to smallholder farmers. Integrating a holistic and pipe geometries—to provide a versatile design tool for cost-
system model with a cost-optimization scheme can enable the optimized, solar-powered drip irrigation systems.
design and implementation of low-cost, solar-powered drip
irrigations systems, ultimately making this technology more cost- NOMENCLATURE
effective for smallholder farmers. 𝐸𝑇0 reference evapotranspiration [mm/day]
This paper describes the algorithms comprising an 𝐸𝑇𝑐 crop-specific evapotranspiration [mm/day]
integrated model of solar-powered drip irrigation systems, 𝐾𝑐 crop coefficient
consisting of agronomic, hydraulic, pump, and power system 𝐾𝑐,𝑖𝑛𝑖/𝑚𝑖𝑑/𝑒𝑛𝑑 crop coefficient during initial/middle/late-season
modules. It also introduces a preliminary optimization scheme growth stages
for the power system, which uses the system hydraulics and pump 𝑅𝑛𝑒𝑡 net radiation at the crop surface [MJ/m2/day]
curve to select an optimal solar array and energy storage 𝐺 soil heat flux density [MJ/m2/day]
configuration that minimizes capital cost. 𝑇 air temperature [°C]
The system model and power system optimization is applied 𝑢2 wind speed at 2 m height [m/s]
to three case studies, and the resulting power system 𝑒𝑠 saturation vapor pressure [kPa]
configurations are compared to outputs from commercially- 𝑒𝑎 actual vapor pressure [kPa]
available software for sizing solar pumping systems. The results ∆ slope of vapor pressure curve [kPa/°C]
show that the model successfully captures the nuances in crop 𝛾 psychrometric constant [kPa/°C]
type, local weather patterns, and hydraulic system layout 𝐶 time step constant in Penman-Monteith equation
between different cases. This offers a greater level of flexibility 𝐴𝑠𝑢𝑏 area of subunit [m2]

1 Copyright © 2018 ASME

 𝑉𝑠𝑢𝑏 volume of water required by subunit [m3/day] especially in areas where costly off-grid power components,
𝑓𝑤 soil wetted fraction such as solar panels and batteries, are necessary. For a direct-
P pressure [Pa] drive solar-powered system, the pump and photovoltaic array

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

Q flow rate [m3/s] can make up 60% to 80% of the total irrigation system cost [8].
V mean fluid velocity [m/s] Lowering the power system costs would greatly reduce the total
L pipe length [m] system cost, and could make drip irrigation accessible to
𝜌 fluid density [kg/m3] smallholder farmers.
𝑓𝑑 Darcy friction factor Both the pump and power system costs are sensitive to a
D pipe inner diameter [m] complex set of co-dependent factors in the drip irrigation system.
𝜀 pipe roughness [m] For a given geographic location and crop type, agronomy
𝜂𝑃𝑉 efficiency of the solar panels parameters and local weather determine the water demand of the
𝜂𝑃𝑉,𝑛𝑜𝑚 nominal efficiency of the solar panels plant, which inform the hydraulic system design and pump
𝛼𝑃 temperature coefficient [%/°C] selection. This in turn dictates the power system requirements.
𝑇𝑎𝑚𝑏 ambient temperature [°C] Modeling these components as an integrated system can
𝑘 Ross coefficient [°C m2/W] elucidate tradeoffs that cannot be exploited if the components are
𝐺𝐻𝐼 global horizontal irradiance [W/m2] optimized independent of one another.
𝑇𝑠𝑡𝑑 standard testing temperature [°C] Existing software packages tend to focus on modeling these
𝑃𝑃𝑉 solar panel power [W] subsystems independently. For example, EPANet [9] and
𝐴𝑃𝑉 solar panel area [m2] HydrauliCAD [10] enable the simulation of flows and pressures
𝑆𝑂𝐶 state of charge of the battery [J] for a user-defined hydraulic network, but do not offer the ability
𝑆𝑂𝐼 state of irrigation (water delivered) [m3] to simulate power systems or programmatically optimize the
𝑄𝑠𝑦𝑠 system flow rate [m3/s] system layout. Software specifically aimed at irrigation design,
such as IrriCAD [11] or IrriPro [12], have a similar functionality,
𝛥𝑡 time interval [s]
including databases of commercial irrigation components. They
𝑃𝑝𝑢𝑚𝑝 pump power [W]
can simulate drip irrigation system hydraulics, but cannot link
𝜂𝑝𝑢𝑚𝑝 pump efficiency [%] them to the crops’ weather-dependent water demands, or
𝜂𝑏𝑎𝑡𝑡 battery efficiency [%] simulate pump operation over time. Existing solar pump sizing
𝜂𝑐𝑜𝑛𝑣 converter efficiency [%] software, such as Compass from Lorentz and Grundfos pump
sizing software tends to be for wider and larger applications, and
does not have the ability to capture information like the variation
1. INTRODUCTION in water demand for different crop types and weather patterns
This paper presents the development of a system model for [13-14]. The lack of tools for simulating the interconnected
designing and optimizing solar-powered drip irrigation systems nature of the components for an off-grid drip irrigation system
for smallholder farmers. A smallholder farmer works less than limits the ability to optimize the system as a whole.
two hectares of land, yet they collectively produce around 80% The goal of the integrated system optimization is to reduce
of the food consumed in Asia and sub-Saharan Africa [1]. capital cost, which smallholder farmers tend to be sensitive to,
Traditional methods of irrigation, such as flood irrigation and as well as determine key trade-offs between capital cost,
furrow irrigation, are widely used by smallholder farmers, but component quality, and the lifetime production of the system in
these methods waste water and are less productive than other order to optimize the lifecycle cost of the system. A model that
irrigation methods. Drip irrigation delivers a controlled volume can exploit these tradeoffs to reduce the overall system cost
of water directly to the root-zone of the crop and allows farmers could be used by irrigation engineers to design accessible drip
to grow a higher quantity and wider variety of crops compared irrigation systems for smallholder farmers.
to flood irrigation. It can reduce water consumption per unit area This paper describes a system model that has been
by up to 70% and, depending on the crop type, can increase crop developed to capture the behavior of the various subsystems, and
yields by 20-90% compared to flood irrigation [2-6]. which builds the foundation for a full system-level optimization.
Despite the benefits of drip irrigation, the high capital cost It describes a first-generation optimization scheme that selects
of pumps and power systems make it difficult for smallholder the most cost-effective power system for a given set of emitter,
farmers to adopt the technology [7]. The high cost of pumps is hydraulic, and pump parameters.
driven by the fact that traditional pressure-compensating drip
emitters require a minimum pressure of 0.5-1 bar to operate at 2. MODEL FORMULATION
the nominal flow rate, and this requires significant pumping The integrated system model consists of multiple
power. This paper considers a drip system with custom, low- components. The irradiation and weather patterns at a specific
pressure emitters that exhibit pressure compensating behavior at location drive the plant evapotranspiration rate, or the rate at
a minimum pressure of 0.15 bar. This low activation pressure which water evaporates off the plant leaves and soil surface. The
significantly reduces the required pumping power for the drip rate of evapotranspiration drives the water demand of the system,
system [8]. Nevertheless, the power system can be expensive, which in turn defines the pump requirements. The pump

2 Copyright © 2018 ASME

 demand over a longer time, either manually or in direct response
to changes in solar irradiation [15]. The current version of the
system model can compare the impact of irrigation frequency

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

(daily or less frequent) on the power system size and cost. A more
frequent schedule typically lowers the power system size, since
the water demand for each event is reduced. In future generations
of the model, flexible scheduling on an hourly basis will be
implemented to further reduce the overall power system size.
The system model is divided into four modules that capture
the behavior of each system component shown in Fig. 1. Each
module takes in a set of inputs, runs calculations, and outputs
quantities that become part of the input set for the subsequent
modules (Fig. 2). The following parameters are fixed across all
four modules: location, crop type, crop spacing, and physical
constants, such as the gravitational constant and the density of
Figure 1: System model overview showing the interdependence of
water.
the components: irradiation, weather, evapotranspiration rate (ET), The agronomics module takes local crop and weather data
water demand, pump, and power system. for a typical meteorological year as inputs and calculates a daily
evapotranspiration rate (ETc) for the crop at the specified
location. The daily ETc is used to calculate the water demand of
specifications drive the system power requirements, including
the crop throughout the year. The hydraulics module takes the
battery capacity and solar panel area. The size of the power
system geometry, crop spacing, emitter flow rate, and physical
system depends not only on the pump requirements, but also on
constants as inputs, and runs an iterative fluid network
the weather patterns and available irradiation.
calculation to determine the pressure drop throughout the
The model can also exploit the time-variant nature of
system. The pump module takes the pump curve and efficiency
irradiation and evapotranspiration. Flexible scheduling has been
of a specified pump, the water demand calculated in the
shown to reduce power system sizes and costs in off-grid small
agronomics module, and a specified irrigation schedule, and uses
scale water treatment systems by spreading out the power

Figure 2: System model architecture showing functionality of current generation (Gen1) and future generations (Gen2, Gen3). Each column
represents a module in the system model with user inputs (rectangles), code (ovals), and code outputs (yellow rectangles). All generations include
all four modules, but only the code outputs with the appropriate dashed box are allowed to vary in the optimization scheme. Each generation will
increase the model flexibility to allow for broader and more detailed design applications.

3 Copyright © 2018 ASME

it to determine the operating point of the pump for the given humidity (RH), wind speed, and total solar shortwave radiation
system. The power system module takes inputs from the three need to be provided in the weather file.
previous modules to design the most cost-effective power system In the current model implementation, typical meteorological

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

configuration. This module takes irradiation data from the year weather files (ASHRAE IWEC2) are used for both
weather file, calculated water demand from the agronomics evapotranspiration and solar power calculation (Section 2.3).
module, pumping power for the selected operation point from the These weather files were chosen as they give representative
pump module, and component parameters for the solar panels hourly average weather values based on weather data collected
and batteries as inputs. Each of the modules is described in more for that location over the past 20 years.
detail in sections 2.1-2.3. Hourly weather data are adapted to the daily
The full system model is used as the foundation for a evapotranspiration calculation (Eq. 1) as follows. Vapor pressure
preliminary optimization scheme for the power system. In the and the slope of vapor pressure curve are computed using
current version of the model (Generation 1), the emitter design, minimum and maximum hourly temperatures and relative
hydraulic system geometries, pump characteristics, irrigation humidity levels recorded throughout the day. Net radiation at the
schedule and solar panel coefficients are determined by crop surface (𝑅𝑛𝑒𝑡 ) is computed as described in [16], with the
calculations or inputs in previous modules before the power incoming shortwave radiation taken directly from the weather
system optimization begins. The optimization is performed as an data file. Shortwave radiation is summed over 24 hours, and
exhaustive search that iterates through all permutations of panel wind speed is averaged. The case studies in this paper use daily
areas and battery capacities. The algorithm then selects the ET0 only.
power system configurations that meet the desired reliability
metric and plots them against a set of capital cost curves. This 2.1.2 Crop evapotranspiration
enables the selection of a cost-optimized power system Once the reference evapotranspiration is calculated, it is
configuration for a given hydraulic system and pump. Future scaled to the crop-specific evapotranspiration (ETc) using the
generations of optimization algorithms (Generation 2 and 3), single crop coefficient model (Kc):
increasing sequentially in complexity, will be developed with the
aim of creating a flexible design tool for efficient, cost-optimized 𝐸𝑇𝑐 = 𝐾𝑐 𝐸𝑇0 (2)
solar-powered drip irrigation systems. These future capabilities
are outlined in section 4. 𝐾𝑐 is a function of the crop and the stage in the crop’s
development. A crop database was constructed from data in [16],
2.1 Agronomics Module with lengths of development stages and Kc values corresponding
to each stage (𝐾𝑐,𝑖𝑛𝑖 , 𝐾𝑐,𝑚𝑖𝑑 , 𝐾𝑐,𝑒𝑛𝑑 ). Initial and mid-season
2.1.1 Reference evapotranspiration stages have constant coefficients (𝐾𝑐,𝑖𝑛𝑖 and 𝐾𝑐,𝑚𝑖𝑑 ), with a linear
The agronomics module calculates the crop water change in coefficient during the crop development stage
requirement on a daily basis using the method presented in “FAO (between 𝐾𝑐,𝑖𝑛𝑖 and 𝐾𝑐,𝑚𝑖𝑑 ) and late season (between 𝐾𝑐,𝑚𝑖𝑑 and
Irrigation and Drainage Paper 56” [16]. First, the Penman- 𝐾𝑐,𝑒𝑛𝑑 ). When climate conditions are different from the climates
Monteith equation (Eq. 1) is used to calculate the reference used for tabulated Kc values (climates where average
evapotranspiration (ET0) based on meteorological data: 𝑅𝐻𝑚𝑖𝑛,𝑚𝑒𝑎𝑛 ≠ 45% or 𝑢2,𝑚𝑒𝑎𝑛 ≠ 2.0 m/s), 𝐾𝑐,𝑚𝑖𝑑 and 𝐾𝑐,𝑒𝑛𝑑 are
𝐶
adjusted per the equation below, provided that the tabulated
0.408∆(𝑅𝑛𝑒𝑡 −𝐺)+𝛾 𝑢 (𝑒 −𝑒 )
𝐸𝑇0 = 𝑇+273 2 𝑠 𝑎
(1) value of 𝐾𝑐,𝑒𝑛𝑑 (𝑇𝑎𝑏) ≥ 0.45:
∆+𝛾(1+0.34𝑢2)

𝐾𝑐,𝑚𝑖𝑑/𝑒𝑛𝑑 = 𝐾𝑐,𝑚𝑖𝑑/𝑒𝑛𝑑 (𝑇𝑎𝑏) + (0.04(𝑢2,𝑚𝑒𝑎𝑛 − 2) −

Where 𝐸𝑇0 is the reference evapotranspiration [mm/day] for a
ℎ 0.3
grass reference crop of 0.12 m height, 𝑅𝑛𝑒𝑡 is the net radiation at 0.004(𝑅𝐻𝑚𝑖𝑛,𝑚𝑒𝑎𝑛 − 45)) ( ) (3)
3
the crop surface [MJ/m2day], 𝐺 is the soil heat flux density
[MJ/m2day], assumed 0 for daily intervals, 𝑇 is the air where 𝐾𝑐,𝑚𝑖𝑑/𝑒𝑛𝑑 (𝑇𝑎𝑏) is the tabulated value for 𝐾𝑐,𝑚𝑖𝑑 or 𝐾𝑐,𝑒𝑛𝑑 ,
temperature at 2 m height [°C], daily or hourly average, 𝑢2 is the
ℎ is the mean plant height during the mid/late-season stage [m]
wind speed at 2 m height [m/s], 𝑒𝑠 is the saturation vapor
(tabulated), and mean values of 𝑢2,𝑚𝑒𝑎𝑛 and 𝑅𝐻𝑚𝑖𝑛,𝑚𝑒𝑎𝑛 are
pressure [kPa] at air temperature T, 𝑒𝑎 is the actual vapor
calculated for the corresponding growth stage (mid- or late-
pressure [kPa], ∆ is the slope of vapor pressure curve [kPa/°C] at
season). Custom crop coefficients and development stage lengths
air temperature T, 𝛾 = 0.665 × 10−3 is the psychrometric
can be entered by the user if more accurate local data are
constant [kPa/°C], 𝐶 = 900 is the time step constant for daily
available.
intervals, and 𝑃 is the atmospheric pressure [kPa].
Details of the calculations of specific terms in the Penman-
2.1.3. Volumetric water requirement
Monteith equation can be found in [16], Chapter 3.
After ETc is calculated for the given crop, it is converted to
Meteorological variables are extracted directly from weather the volume of water that needs to be delivered to each subunit in
data whenever possible. At minimum, temperature, relative
the field. A subunit is defined as an area irrigated by one submain

4 Copyright © 2018 ASME

with its connected laterals. For a subunit with a given area, 𝐴𝑠𝑢𝑏 , The final output of this module is a system curve that plots
the water volume, 𝑉𝑠𝑢𝑏 [m3/day], is calculated as: the system pressure versus flow rate (Fig. 4b). Due to the
pressure-compensating behavior of the emitters, the system

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

𝑉𝑠𝑢𝑏 = 𝑓𝑤 𝐴𝑠𝑢𝑏 𝐸𝑇𝑐 /1000 (4) curves have a gradual positive slope until the activation pressure
of the emitters, at which point the plot becomes nearly vertical.
where 𝑓𝑤 is the soil wetted fraction, which is assumed to be 0.3 The ideal pump operation point is at the activation pressure, or
for drip irrigation [16]. the point where the slope of the system curve changes. The left-
most point on the curve is where the pressure losses from the
2.2 Hydraulics and Pump Module filters and pipe network equal the inlet pressure head. The model
The hydraulics module accepts user inputs on pipe lengths, calculates zero flow at this point, but in reality, operating at this
inner diameters, pipe and emitter spacings, and emitter point would probably result in some flow at the emitters closest
pressure-flow characteristics, and calculates the pressures and to the submain and no flow at the emitters further along the
flows at all points in the system. When this calculation is run laterals (Fig. 4a). The pump module plots the pump curve against
for a range of input pressures, a system curve of flow rate the system curve and calculates the intersection point to
versus input pressure is generated. The module assumes that a determine the operating point of the given pump.
water source is connected to one main pipe, which serves one
or more submains. Each submain delivers water to multiple
laterals. Each lateral has a number of emitters, assumed to be
equally spaced. The system layout is generated automatically
based on the specified pipe dimensions. An example layout is
shown in Fig. 4(a).
For each pipe, the user specifies the length, inner diameter,
roughness, and the spacing of smaller pipes or emitters
connected to the pipe. Flow out of drip emitters is modeled as
linear when 0 < 𝑃 < 𝑃𝑎𝑐𝑡 , and as following the curve 𝑄 = 𝑘𝑃 𝑥
when 𝑃 ≥ 𝑃𝑎𝑐𝑡 . Here, 𝑘 is the flow coefficient, 𝑥 is the pressure
compensation exponent, and 𝑃𝑎𝑐𝑡 is the activation pressure,
defined as the pressure at which the emitter begins exhibiting
pressure compensating behavior.
The calculation of flow and pressures at every point in the
system is performed iteratively, following the flow chart in Fig.
3. Iterations are run until convergence in flow rates, occurs with
a maximum convergence error of 1 L/h for total flow entering
any submain. Major pressure losses are calculated using the
Darcy-Weisbach equation (Eq. 5). Minor pressure losses are
modeled for tee fittings at the start of each submain and lateral
pipe (Eq. 6), where K = 1 for tee losses in branch flow [17].

𝐿 𝜌𝑉 2
Figure 3: Flow chart of iterative hydraulic module algorithm. The
∆𝑃𝑚𝑎𝑗𝑜𝑟 = 𝑓𝑑 (5) pressure and flow rate at each emitter in the system is calculated by
𝐷 2
𝜌𝑉 2 initializing all emitters to their rated flow rate, computing pressure
∆𝑃𝑚𝑖𝑛𝑜𝑟 = 𝐾 (6) drops in all segments under that assumption, re-computing the flow
2
rates, and iterating in this manner until the flow rates at the inlet of
The Swamee-Jain formula is used for the Darcy friction factor 𝑓𝑑 each lateral and submain converge.
in turbulent flow [18]:
64
2.3 Power System Module and Gen1 Optimization
𝑓𝑑 = for 𝑅𝑒 < 2300 (7) The power system module takes inputs from the other
𝑅𝑒
𝜀 5.74 modules, including the time-variant water demand from the
𝑓𝑑 = 0.25 [𝑙𝑜𝑔10 ( + ) ]−2 for 𝑅𝑒 ≥ 2300 (8)
3.7𝐷 𝑅𝑒 0.9 agronomics module, hourly solar irradiance and temperature
from the weather file, and the pumping power requirements at
In Eq. 7 and 8, Re is the Reynolds number, 𝑓𝑑 is the Darcy the operation point from the pump module. It also takes as inputs
friction factor, D is the pipe inner diameter [m], and 𝜀 is the pipe an array of panel areas, battery capacities, and the associated
roughness [m]. Pressure losses for water flowing over an emitter panel and battery electrical efficiencies and capital costs. The
are neglected. Pressure losses due to filters and other fittings are model then iterates through all possible panel area and battery
added as a constant 0.3 bar, representing the expected worst-case capacity combinations and outputs the associated power system
pressure drop. Only steady state operation is simulated. cost along with the number of days the water demand was not

5 Copyright © 2018 ASME

met in an irrigation season, or number of “failure” days. In this The current power flow logic in the module (Fig. 5)
way, the power system design space is explored and the lowest considers the irradiation at each time interval throughout the day,
cost reliable power system can be found. and calculates the state of charge (SOC) of the battery and the

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

Hourly irradiation and temperature data and constants from state of irrigation (SOI), or water delivered, at each interval.
the PV panel datasheets are used to calculate the efficiency of the First, the module checks if the water demand has been met. If the
solar panels at each time interval, 𝜂𝑃𝑉 (𝑡): water demand for the day has not been met, it checks if there is
enough irradiation in the time interval to power the pump. If
𝜂𝑃𝑉 (𝑡) = 𝜂𝑃𝑉,𝑛𝑜𝑚 ∙ (1 + 𝛼𝑃 ∙ (𝑇𝑎𝑚𝑏 (𝑡) + 𝑘 ∙ 𝐺𝐻𝐼(𝑡) − 𝑇𝑠𝑡𝑑 )) (9) there is enough irradiation, it powers the pump with PV power
and uses any extra PV power to charge the battery. This adds to
where 𝜂𝑃𝑉,𝑛𝑜𝑚 , the nominal efficiency of the panels, 𝛼𝑃 , the the SOI and SOC as shown in Eq. 11 and 12:
temperature coefficient [%/°C], and 𝑇𝑠𝑡𝑑 , the standard testing
temperature [°C] are all constants from the panel datasheet; 𝑆𝑂𝐼(𝑡) = 𝑄𝑠𝑦𝑠 ∙ 𝛥𝑡 (11)
𝑇𝑎𝑚𝑏 (𝑡), the ambient temperature [°C], and 𝐺𝐻𝐼(𝑡), the global 𝑆𝑂𝐶(𝑡) = 𝑆𝑂𝐶(𝑡 − 1) + 𝛥𝑡 ∙ (𝑃𝑃𝑉 (𝑡) −
𝑃𝑝𝑢𝑚𝑝
) ∙ 𝜂𝑏𝑎𝑡𝑡 ∙ 𝜂𝑐𝑜𝑛𝑣 (12)
horizontal irradiance [W/m2], are taken from the local weather 𝜂𝑝𝑢𝑚𝑝

data. The Ross coefficient, k, relates irradiance to module

temperature (k = 0.025 °C 𝑚2 /W). where 𝑄𝑠𝑦𝑠 is the system flow rate in 𝑚3 per second, 𝛥𝑡 is the
The PV power [W] at each iteration and time interval, time interval in seconds, 𝑃𝑝𝑢𝑚𝑝 is the pump power in Watts,
𝑃𝑃𝑉 (𝑡), is calculated in Eq. 10, with 𝐴𝑃𝑉 as the PV panel area 𝜂𝑝𝑢𝑚𝑝 is the pump efficiency, 𝜂𝑏𝑎𝑡𝑡 is the battery efficiency, and
[𝑚2 ] at a specified iteration. 𝜂𝑐𝑜𝑛𝑣 is the converter efficiency. For this paper, 𝛥𝑡 = 3600 s,
𝜂𝑏𝑎𝑡𝑡 = 85%, and 𝜂𝑐𝑜𝑛𝑣 = 95%.
𝑃𝑃𝑉 (𝑡) = 𝜂𝑃𝑉 (𝑡) ∙ 𝐴𝑃𝑉 ∙ 𝐺𝐻𝐼(𝑡) (10) If there is not enough irradiation to power the pump directly,
the module checks if there is enough battery capacity to run the
pump for the time interval 𝛥𝑡. If there is enough battery capacity,
it powers the pump with the battery, which adds to SOI and
subtracts from SOC as shown in Eq. 11 and 13:
𝑃𝑝𝑢𝑚𝑝
𝑆𝑂𝐶(𝑡) = 𝑆𝑂𝐶(𝑡 − 1) − 𝛥𝑡 ∙ ( ) (13)
𝜂𝑝𝑢𝑚𝑝 ∙𝜂𝑐𝑜𝑛𝑣

If there is not enough battery capacity to power the pump, the

SOI is set to zero (Eq. 14) for the time interval. If there is
available irradiation, the panels charge the battery, adding to the
SOC (Eq. 15). If there is insufficient irradiation, the SOC
remains constant for this time interval.

𝑆𝑂𝐼(𝑡) = 0 (14)
𝑆𝑂𝐶(𝑡) = 𝑆𝑂𝐶(𝑡 − 1) + 𝛥𝑡 ∙ (𝑃𝑃𝑉 (𝑛, 𝑡) ∙ 𝜂𝑏𝑎𝑡𝑡 ∙ 𝜂𝑐𝑜𝑛𝑣 ) (15)

Figure 4: System layout and system curves calculated by the

hydraulics module for a sample field. (a) System layout showing
the pipe network; (b) Pressure-flow curves for the system and the
selected pump. Figure 5: Power flow logic used in the power system module to
model direct drive and PV-battery power system configurations.

6 Copyright © 2018 ASME

If the water demand is met, the SOI is set to zero (Eq. 14) for the panels in Morocco cost US$113 per square meter, the panels in
time interval, and the module checks if the battery is full. If it is Jordan cost $103 per square meter, and the batteries in both
not full, the panels charge the battery, which increases the SOC countries cost $98.70 per MJ. The pump selected for all three

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

(Eq. 15). If it is full, the SOC is set to its value from the previous cases is a Lorentz CS-F4-3 0.7 kW surface pump because it was
time step. powerful enough to meet the operating point at all three sites.
The SOI is summed over the entire day to determine the total The power system configurations were designed to be 100%
volume of water delivered, which is compared to that day’s water reliable during the irrigation season, which is assumed to last
demand. If the SOI sum does not meet the water demand, a from March 1st to December 15th. Reliability is defined as the
failure is recorded for that day. There is also a set of limits such ability of the system to meet the crop water demand every day of
that the battery cannot be charged beyond its maximum capacity the irrigation season in a typical meteorological year. As the
or drained below 50% of its capacity, for durability reasons. The actual weather patterns and irrigation season vary from year to
power system module helps assess the trade-off between using a year, the resulting design may not be 100% reliable for any given
direct-drive system versus a PV system with batteries capable of calendar year.
storing energy during times of high irradiance and releasing it The final output of the simulation is a cost curve plot that
during times of low irradiance (Fig. 6). Section 3.1 explores this indicates the price of various power system configurations for
trade-off further by analyzing several test cases with varying each design case. Each power system configuration consists of a
field sizes, water demands, and component costs. panel area and battery capacity, and is evaluated by its capital
cost only. The two points of interest for each case are the cost-
optimized, or least expensive, configuration and the direct drive
configuration, which is the smallest reliable configuration where
only solar panels are needed to power the pump. This second
point is used to compare the design to the industry standard,
which only uses the direct drive configurations for systems of
this scale.
For the young citrus orchard in Sharhabeel (Jordan), the
least expensive power system has a panel area of 1.5 m2, a 360
Wh battery capacity, and a capital cost of $282 (Fig. 7). The
direct drive configuration requires a 5 m2 panel area at a capital
cost of $515. For the Saada (Morocco) young olive grove, the
least expensive configuration has a 1.5 m2 panel area, a 360 Wh
battery capacity, and a capital cost of $297 (Fig. 8). The direct
drive configuration requires a 3 m2 panel area at $339. For the
Figure 6: Example of available solar power (blue dashed line) and
system power requirement (red solid line) over time, highlighting times
Beni Mellal (Morocco) mature citrus orchard, the least expensive
of battery charging (green solid circles), when solar power exceeds configuration has a 5.5 m2 panel area, a 960 Wh battery capacity,
demand, and battery discharging (orange circles) when solar power is and a capital cost of $963 (Fig. 9). The direct drive configuration
less than demand. requires a 10.5 m2 panel area at $1,187.
As expected, the size and cost of the power system increases
3.1 MODEL RESULTS with the field area. However, these results also highlight the
An initial assessment of the system model was performed complex interdependence of various system elements captured
by running simulations for three design cases with different by the system model. An example is the Sharhabeel and Saada
locations, crop types and field sizes. These results were direct drive cases. The direct drive configuration for Sharhabeel
compared to results from a commercial design software for solar requires an additional two square meters of panel area and costs
pumping systems that was used as an industry standard almost $180 more than the configuration for Saada. However,
benchmark. the normalized solar panel cost in Jordan (where Shahabeel is
The three cases were a young citrus orchard in Sharhabeel located) is $103 per m2, lower than the $113 per m2 cost in
in the Jordan Valley, a young olive grove in Saada near Morocco (where Saada is located). In addition, the Saada olive
Marrakech, Morocco, and a mature citrus orchard in Beni Mellal, grove is over twice the size of the Sharhabeel citrus orchard. This
Morocco. The site specifications used in the simulation are discrepancy is due to several factors throughout the system that
shown in Table 1. The “young” and “mature” distinction is impact the power requirements. The pipe diameters in
reflected in the water demand calculation, as more mature trees Sharhabeel are smaller than those in Saada, which contributes to
tend to need more water. Flexible scheduling was implemented hydraulic losses and drives up the power required to operate the
in these case studies by switching irrigation to a daily schedule system. According to the typical solar irradiance data used in the
from a Monday-Wednesday-Friday schedule, which is a model, irradiance in Saada tends to be higher than in Sharhabeel
common irrigation schedule for non-water intensive crops. All (Fig. 10).
three cases were simulated using pricing determined by quotes
from local contractors for the solar panels and batteries. The

7 Copyright © 2018 ASME

 Table 1: Specifications used for each case in the simulations.

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

Site Sharhabeel (Jordan) Saada (Morocco) Beni Mellal (Morocco)
Area [ha] 0.2 0.52 1.1
Crop type Citrus, young Olive, young Citrus, mature
Weather station location Irbid, Jordan Marrakech, Morocco Marrakech, Morocco
Pipe length [m]
90.5, 45, 45 300, 82, 72 760, 237, 49
(main, submain, lateral)
Pipe inner diameter [mm]
44.2, 44.2, 14.4 83, 59, 16 66.4, 55.8, 14.4
(main, submain, lateral)
Num. emitters 320 360 676
Fluid network component head
3.1 3.1 3.6
loss [m]
Estimated NPSH [m] 1.5 1.5 2
Irrigation schedule Daily Daily Daily
Pump model Lorentz CS-F4-3 Lorentz CS-F4-3 Lorentz CS-F4-3
PV panel price [US$/m2] 103 113 113
Battery price [US$/MJ] 98.7 98.7 98.7

Figure 7: Power system configurations for a young citrus orchard Figure 8: Power system configurations for a young olive grove in
in Sharhabeel, Jordan. The least expensive configuration has panel Saada, Morocco. The least expensive configuration has panel area
area of 1.5 m2, a 360 Wh battery capacity, and a capital cost of $282. of 1.5 m2, a 360 Wh battery capacity, and a capital cost of $297. The
The direct drive configuration requires a panel area of 5 m2 at $515. direct drive configuration requires a panel area of 3 m2 at $339.

Together, these factors result in a larger and more expensive The model does have some shortcomings that will need to
direct drive power system for Sharhabeel, despite the Saada site be addressed in future implementations. For example, the Beni
being twice as large in size and located in a region with more Mellal case indicates that a direct drive system would cost over
expensive solar panels. This comparison demonstrates the $200 more than the cost-optimized configuration and require
capacity of the system model to capture the behavior and almost double the panel area. However, these price points are
complex interactions of the system components discussed in misleading without considering the lifecycle costs associated
section 2. with panels and batteries. Lifecycle costs generally include
maintenance, replacement, and financing interest costs for all the

8 Copyright © 2018 ASME

 Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020
Figure 9: Power system configurations for a mature citrus orchard Figure 10: Daily solar irradiance in Sharhabeel and Saada for a
in Beni Mellal, Morocco. The least expensive configuration has 5.5 typical meteorological year. The solar irradiance available in Saada
m2 panel area, a 960 Wh battery capacity and a capital cost of $963. tends to be higher than in Sharhabeel.
The direct drive configuration requires a panel area of 10.5 m2 at
$1,187.

system components [19]. Batteries tend to have a significant This model will take more inputs from the PV datasheets and
lifecycle cost because they need to be replaced regularly. The accurately characterize the maximum power point for various
panels selected for these test cases have a lifetime of irradiance and temperature conditions.
approximately 20 years [20-21]. Assuming the lifetime of the
power system is equal to that of the panels, the batteries in the 3.2 COMPARISON TO INDUSTRY STANDARD
lowest capital cost configurations for Sharhabeel, Saada and The system model is compared to a commercially available
Beni Mellal would have to be replaced 25, 8, and 6 times, solar pump sizing software called Compass [13]. Compass takes
respectively. Using the quoted battery cost, and assuming the inputs of location (latitude, longitude, and altitude), total
batteries have a lifetime of 300 charging cycles, this would dynamic pressure head, pipe length, diameter and pipe material,
correspond to an additional $3,198 for Sharhabeel, $1,023 for and a fixed daily water demand. The software outputs the solar
Saada, and $2,365 for Beni Mellal in battery replacement costs pump and panel configurations that would meet these
over the lifetime of the power system. This means the battery requirements from a database of Lorentz solar pumps and
cost is artificially low in the current model, making a direct drive Lorentz or user-input panels. It also outputs the optimal
configuration appear impractical for a site like Beni Mellal. In operating point, pressure and flow rate for the pump, and does
reality, systems with batteries may become more expensive than not allow for constant flow rate systems. The software does
direct drive systems when lifecycle costs are taken into account. consider sizing for various solar conditions: an average solar
Another shortcoming is the accuracy of the solar panel month, a month with the most or least solar output, a custom
model. The model estimates the degrading effects of temperature season (various months can be selected), a defined kWh/m2 solar
and low solar irradiance on panel power output using Eq. 9-10. day, or the driest month. It should be noted that Compass only
A more accurate model for characterizing PV power that will be considers direct drive systems, so the Compass results were
implemented in future generations is the single diode model.
Table 2: Comparison of Compass and model results for direct drive panel area calculation.
Pressure Max. daily Compass calc. Compass reliability Model calc. panel Model Reliability
Location
head [m] water [m3] panel area [m2]¹ (# failure days) area [m2]¹ (# failure days)
Sharhabeel, Jordan 7.5 5.0 3.9 1 5.0 0
Saada, Morocco 6.5 16 4.8 0 3.0 0
Beni Mellal, Morocco 10 19 4.8 79 10.5 0

¹The resolution of the system model panel area calculation is 0.5 m2, whereas the resolution of the Compass calculation is a unit panel area,
which is 2 m2 for the Jordan case and 1.6 m2 for the Morocco cases.

9 Copyright © 2018 ASME

compared to the direct drive configurations from the model oversized for Saada and undersized for Beni Mellal. The system
results. model is more adaptable as it allows for different panel areas to
The results from the system model for the three case studies be considered for a given pump and panel model. The system

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

are compared to results from Compass for the same three cases, model calculated different panel areas for the Saada and Beni
considering an average solar month for sizing (Table 2). The Mellal cases and therefore is able to design customized systems
same pump that is used in the system model, the Lorentz CS-F4- for each case.
3 0.7 kW surface pump, is used for each of the three cases in There are additional benefits to the system model over
Compass. The same panel parameters that are used in the system Compass, such as its specificity to irrigation system design and
calculations are used in Compass: a 270 Wp 1.6 m2 panel for the its ability to consider batteries in addition to direct drive power
Morocco cases and a 300 Wp 2 m2 panel for the Jordan case. The systems. The system model allows a user to consider constant
compared results are the required solar panel area and the flow and constant pressure systems, which has direct
reliability of the power system designs. Reliability is quantified applications to drip irrigation with pressure compensating
as the number of days the system design failed to deliver the emitters. Furthermore, the system model allows for a wider
amount of water required over the irrigation season. The number range of inputs compared to Compass, such as hourly irradiance
of failure days output by both Compass and the system model and daily water demand.
were calculated using the system model. A 100% reliable design Compass has a number of advantages as well, including a
is one in which the number of failure days is zero, as described descriptive pump and solar panel database and an interactive user
in section 3.1. interface, which may inform future generations of the system
Table 2 shows that the Compass and system model results model. The comparison of the system model to Compass
are not consistent. For Sharhabeel, Compass calculated a 20% indicates that there is a need for a more detailed design software
smaller panel area compared to the system model, producing a than what currently exists in the industry to accurately
design that failed for 1 day during the irrigation season. This characterize the cost and reliability of a solar-powered drip
shows that if the reliability limit is relaxed slightly, the optimized irrigation system.
designs could be smaller and cheaper. For Saada, Compass
calculated a 60% higher panel area, which would be more 4. CONCLUSIONS & FUTURE WORK
expensive, yet just as reliable compared to the cheaper system Designing and cost-optimizing solar-powered drip irrigation
model calculation. For Beni Mellal, Compass calculated a 54% systems requires a system model that captures the
lower panel area than the system model, but this design failed for interdependence of the crop water demand, local weather
27% of the irrigation season. Notably, Compass calculated the patterns, hydraulic system, pump, and power system
same panel area for Saada and Beni Mellal, even though Beni configuration. This paper presents a system model that captures
Mellal is a much larger site with a higher operating pressure and these relationships, as well as a first generation power system
daily water requirements. optimization for capital cost. This optimization scheme is used
The discrepancies between the system model and Compass to design power systems for three drip irrigation case study sites,
may come from several sources. Compass assumes that the daily one in Jordan and two in Morocco; the results of which
water output is constant (an average), whereas the system model demonstrate the interconnectedness of the parameters within the
calculates the daily water demand based on the crop model. Case study results are compared to results from an
evapotranspiration, which changes every day based on irradiance industry standard software, Compass. While the industry
and weather patterns. This means that Compass is less accurate software has a much wider range of applications, with a focus on
in calculating the water demand. The software also automatically larger systems, it does not have the sensitivity to design a cost-
selects the optimal flow rate for a given operating pressure and optimized system. The current model offers a level of flexibility
the available solar irradiance, which makes it ill-suited to design that is missing in the Compass software.
constant flow rate systems. In Generation 2 of the system model, the pump
Compass also assumes an average monthly solar irradiance characteristics will become variables and the model will iterate
based on the given location. The system model is fed hourly through a selection of pumps and power system configurations.
irradiance data from the given location, enabling it to account for This generation will also consider variable pump operation.
daily variations and “worst case” days when irradiance is low. Generation 3 of the model will allow the emitter flow rate and
Finally, Compass creates specific system designs from its the hydraulic system geometries and layout to vary. In each
database of pumps and solar panels and then outputs designs that generation of the model, an optimization scheme will find the
meet the user-input specifications. Changing the average daily most cost-effective system configuration for a given location and
water output in Compass does not change the number of panels crop type. The lifecycle costs of these components will be
required by a pump in a specific design, it only changes which considered in the cost optimization along with capital costs, and
pump and panel design combinations are presented by the more accurate component models will be implemented for the
software. This method has shortcomings as, for example, the system components. Hourly ET0 values will be used in future
same design was shown to meet the specifications for both Saada generations of the optimization scheme, and an updated model
and Beni Mellal, despite the fact that Beni Mellal is twice as large of the solar panels that captures their weather-dependent power
as Saada. This means that the Compass design was most likely production will be included. The system model will also

10 Copyright © 2018 ASME

incorporate higher resolution solar irradiance data and consider Season Vegetables Under Plastic Tunnel.” Int. J. Agric. Crop
alternative energy storage methods to batteries, such as water Sci., 7(4), pp. 185-190.
storage tanks at a height. As the model uses average weather data

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

for a typical meteorological year to predict water demand, it may [5] Nkya, K., Mbowe, A., and Makoi, J.H.J.R., 2015, “Low-Cost
be beneficial to define a safety factor or predictive weather Irrigation Technology, in the Context of Sustainable Land
modeling that can take into account the variation in annual Management and Adaptation to Climate Change in the
weather patterns. This could potentially protect crop yields by Kilimanjaro Region.” J. Environ. Earth Sci., 5(7), pp. 45–56.
improving the reliability of the system in any given year.
The three sites used as simulation cases are three of nine [6] Namara, R.E., Upadhyay, B., and Nagar, R.K., 2005,
available field trial sites in Jordan and Morocco where cost- “Adoption and Impacts of Microirrigation Technologies
optimal systems designed using the system model are being Empirical Results from Selected Localities of Maharashtra and
installed. The data collected from these field sites will be used to Gujarat States of India,” International Water Management
validate and improve the system model. The installation of the Institute, Colombo, Sri Lanka, Technical Report No. 93.
first two of these systems is currently in progress at Saada and
Sharhabeel, and field measurements of the pump efficiency, [7] Friedlander, L., Tal, A., and Lazarovitch, N., 2013,
system pressure drops, and system flow rate have been taken at “Technical considerations affecting adoption of drip irrigation in
Sharhabeel. These measurements, along with pressure, flow rate sub-Saharan Africa.” Agricultural Water Management, 126, pp.
and available solar irradiance data, will be incorporated into the 125-132.
model for each case simulation. An on-site weather station will
be installed to collect more accurate, high resolution irradiance, [8] Shamshery, P., and Winter, A.G., 2017, “Shape and Form
wind, temperature, relative humidity, and rainfall data, which Optimization of On-Line Pressure-Compensating Drip Emitters
will also be incorporated into the model. The outcome will be a to Achieve Lower Activation Pressure.” ASME J. Mech. Des.,
validated analytical model that can be used to design solar- 140(3):035001-035001-7. doi:10.1115/1.4038211.
powered drip irrigation systems for a given location and crop
type. This model will be capable of exploiting the [9] United States Environmental Protection Agency, 2017,
interdependence of the hydraulic, pumping, and power systems “EPANET: Application for Modeling Drinking Water
to optimize for the lowest total system cost. Distribution Systems.” From https://www.epa.gov/water-
research/epanet.
ACKNOWLEDGMENTS
The authors would like to acknowledge the staff of Methods [10] HydrauliCAD Software, 2016, “HydrauliCAD.” From
for Irrigation and Agriculture (MIRRA) and the International https://hydraulicad.com/.
Center for Agriculture Research in the Dry Areas (ICARDA) for
their support and advice. The authors would also like to [11] AEI Software, 2018, “IrriCAD: Irrigation Design
acknowledge Estevan Villarreal of Lorentz for his invaluable Software.” From https://www.irricad.com/.
input on solar pumping systems, and Susan Amrose for her help
preparing this publication. [12] Irriworks, 2018, “IrriPro.” From
https://www.irriworks.com/irripro.
REFERENCES
[1] IFAD, 2013, “Smallholders, food security, and the [13] Lorentz, 2018, “Connected: Simple sizing and
environment,” International Fund for Agricultural Development, configuration.” From https://www.lorentz.de/products-and-
Rome, Italy. http://www.ifad.org/climate/resources/smallholder technology/products/connected.
s_report.pdf.
[14] Grundfos, 2018, “Product Selection” From https://product-
[2] Burney, J., Woltering, L., Burke, M., Naylor, R., and selection.grundfos.com/front-page.html?qcid=360685742.
Pasternak, D., 2010, “Solar-powered drip irrigation enhances
food security in the Sudano-Sahel.” Proceedings of the National [15] Watson, S., 2017, “Cost Optimization of a Solar-Powered
Academy of Sciences of the United States of America, 107(5), pp. Electrodialysis Desalination System.” M.S. Thesis.
1848-1853. Massachusetts Institute of Technology, Cambridge, MA.

[3] Sivanappan, R.K., 1994, “Prospects of micro-irrigation in [16] FAO, 1998, “FAO Irrigation and drainage paper 56: Crop
India.” Irrig. Drain. Syst., 8, pp. 49–58. doi: 10.1007/ evapotranspiration - Guidelines for computing crop water
BF00880798. requirements.” Food and Agriculture Organization of the United
Nations, Rome.
[4] Iqbal, M., Sahi, F.-U.-H., Hussain, T., Aadal, N.K., Azeem,
M.T., and Tariq, M., 2014, “Evaluation of Comparative Water [17] White, F.M., 2003, Fluid Mechanics, 5th Ed.
Use Efficiency of Furrow and Drip Irrigation Systems for Off-

11 Copyright © 2018 ASME

[18] Swamee, P.K., and Jain, A.K., 1976, “Explicit equations for
pipe flow problems.” Journal of the Hydraulics Division,
American Society of Civil Engineers, 107(5), pp. 657-664.

Downloaded from http://asmedigitalcollection.asme.org/IDETC-CIE/proceedings-pdf/IDETC-CIE2018/51760/V02BT03A022/2475759/v02bt03a022-detc2018-86297.pdf by Massachusetts Inst Of Tech. user on 13 August 2020

[19] Lai, C.S., and McCulloch, M.D., 2017, “Levelized cost of
electricity for solar photovoltaic and electrical energy storage.”
Applied Energy, 109, pp. 191-203.

[20] Jain Photovoltaic, 2018, “Photovoltaic Module JJ-M672.”

Jain Irrigation Systems Ltd.

[21] Canadian Solar, 2018, “CS6P-260/265/270P.” Canadian

Solar Inc.

12 Copyright © 2018 ASME