Chapter 12Design and Implementation of Off-Grid Systems

CHAPTER 12
DESIGN AND
IMPLEMENTATIO
N OF OFF-GRID
SYSTEMS

Introduction
■ This chapter considers the design and implementation of off-grid
systems
■ Mini-grid design is an exercise in design under constraints. The
constraints most obviously include economic and technical issues,
but often times social, political, legal, and environmental
considerations influence the final design
■ Off-grid system design is almost always multi-objective
■ The systems are often deployed not only to provide electricity access
but also to support development goals such as community resiliency
and empowerment
■ Other off-grid systems are implemented to generate profit
■ This chapter is primarily concerned with mini-grids. However, the
same general concepts apply to off-grid systems serving single users.
Stand-alone systems, such as those serving a single school or
hospital, will have a more simplified distribution system than covered
here.

Mini-grid Life Cycle
■ From the initial decision to develop a mini-grid to the final
decommissioning or expansion, the life cycle is composed of eight
steps:
1. Prospecting and screening
2. Site assessment
3. Decision
4. Technical and commercial design
5. Pre-implementation
6. Implementation
7. Ongoing operation
8. Expansion or retirement

A comment
■ When implementing an off-grid project:
– Never forget that off- grid communities are among the most
underserved and at-risk in the world
– These communities deserve better than to be treated as test
grounds for electricity access projects
– There are probably far more failed or unsustainable off-grid
systems than there are those that are successful
– Just as access to electricity can dramatically improve a
community’s outlook, a poorly planned project can ultimately
cause more harm than good
– The good news is that careful planning and diligence and guard
against dramatic failure can provide the best opportunity for
success

Site Selection
■ In general, and from a techno-economic perspective, communities
that are better candidates for sustainable mini-grids are those that
have some or all of these characteristics:
– the national grid is located far away (20 km or more) with no credible
plans for it to be extended to the community in the next several years
– there are no other off-grid systems installed or planned to be installed
– the estimated demand for electricity is high, and the planned uses are
productive or improve the quality of life
– one or more energy resources are suitable for electricity generation
– there is sufficient ability and willingness to pay for electricity at a cost-
reflective rate or a rate that aligns with the implementing
organization’s ability to operate the mini-grid
– the community is densely populated, which reduces distribution
infrastructure costs
– there is basic infrastructure, including roads and cellular network
coverage
– the community is politically stable
– there is a low risk of theft and vandalism

Site Selection
■ Other characteristics can be important, especially if the primary goal
of the mini- grid is not commercial:
– access to electricity is a high-priority developmental goal for the
community;
– access to electricity aligns with and supports other
developmental goals
■ health care
■ education
– the community is able to organize itself to manage the mini-grid
(if applicable).

Step 1: Prospecting and Screening
■ Resources are limited, and the success of a mini-grid largely
depends on its location
■ The basic goal of this step is to develop a “short list” of communities
where the mini-grid is believed to have best opportunity to be
successful.
■ There may be thousands of off-grid communities in each country or
region. They must be screened quickly and inexpensively. Secondary
data from census or other data set that provides basic demographic
information about communities can be used
■ These data are even more powerful when combined with the
location of existing distribution lines, energy resources, and other
geospatial data
■ Screening is often done based on population density, distance to the
national grid, and perhaps income

Fig. 12.1 Online screening tools combine demographic and geospatial data to
visualize communities where off-grid solutions might be economically viable.
Shown is a screenshot of the Electrification Pathways geospatial tool, which is
part of the ENERGYDATA.INFO and World Bank Group Energy and Extractives
Open Data and Analytics Initiative and created by the KTH University (courtesy
of World Bank Group, KTH University, and the Energy Sector Management

Step 2: Site Assessment
■ At this stage, the organization is investing considerable time and resources evaluating
each site. The basic goal of the assessment is to gather primary data to support
preliminary design and business modeling of the mini-grid. Typical activities include:
– observe the community firsthand;
– meet with local officials—this might involve governmental officers such as the
District Commissioner, as well as traditional leaders such as chiefs, elders, or
community groups;
– confirm that the secondary data used in the prospecting and screening step is
reasonable;
– conduct surveys or focus groups to estimate present fuel consumption and
predict future electric load and consumption patterns;
– check for cellular network coverage. This is important if the mini-grid will use
mobile payments and/or remote data monitoring and control;
– identify existing and future local resources and economic activities;
– collect preliminary data for resource assessment;
– identify locations for mini-grid assets—generation equipment and facilities and
distribution pathways—including discussion with landowners if land must be
leased or purchased.

Step 3: Decision
■ The findings from the assessment trip can further reduce the list of
candidate communities and allow them to be prioritized
■ This might be done from a purely commercial perspective
■ Other tools such as PESTLE (Political, Economic, Social, Technology,
Legal and Environmental) analysis and Risk Matrices can be useful
in ranking the communities
■ Depending on the implementing organization’s resources, perhaps
just one or two communities advance to the next step.

Step 4: Technical Design and Business Plan
■ The technical design details are discussed later
■ An economic evaluation
– capital and financing requirements
– operation and maintenance costs
■ Systems intended to produce a profit or at least break even require a
business plan
■ Most for-profit mini-grids use a metric known as average revenue per
user (ARPU), which is the estimated or actual revenue associated
with each connection over a period of time—typically 1 month

Step 5: Pre-installation
■ Pre-installation activities include securing the required permits and
approvals, may need to begin months in advance of a targeted
installation date
■ Many countries exempt mini-grids under a certain capacity
– for example, 50 kW, from having to obtain a license or permit
■ Be aware that importing materials can be a lengthy process.
■ Community meetings and/or customer orientation like that shown in
next page occur during the pre-installation stage
■ The goal is to inform the community about the pending implementation
■ The electricity tariff is discussed, questions are answered, and feedback
from the community is solicited
■ This often includes basic safety and technical topics including, for
example, the typical cost of running various appliances.

Community meetings, like this one near Siavonga, Zambia,
are important in introducing a project during the pre-
installation stage (courtesy of A. Stewart)

Step 7: Ongoing Operation
■ After installation, the mini-grid begins serving its users
■ Some mini-grids require very little human intervention
– powered by a PV array
■ Others require daily or continuous oversight
– A mini-grid using biomass gasification
■ Some mini-grids require administrative and customer support staff
– to collect payments
– Maintenance routinely and as required
– local people can be trained to perform these tasks

Step 8: Expansion or Retirement
■ Successful mini-grids can be expanded to supply more users or to
generally improve the electricity access tier
– by increasing its availability
■ But when mini-grids are not successful, the operator might decide to
relocate the major assets— the battery bank and PV array, for
example—to another mini-grid, turn the asset over to the community
or other organization, or fully retire and decommission the mini-grid
■ Retiring a mini-grid involves removing its assets, disposing them in
an environmentally responsible way, and returning the land to its
original condition
■ Retirement is not always a negative outcome
– a mini-grid might be retired because the national grid is finally
extended to the community

Modular Design
■ The desire to rapidly scale and reduce design and implementation
costs has pushed some organizations toward using standardized or
modular designs
– a robust portable structure like a shipping container is outfitted
with the energy production components—PV array, batteries,
and controllers—prior to being moved to the community

Energy Conversion Technology Selection
■ Energy conversion technologies in mini-grids are
– conventional or biomass gen sets
– PV arrays
– micro hydro power (MHP)
– wind energy conversion systems (WECS)
■ Each has its own advantages and disadvantages
■ When determining which to use, the designer should consider the
characteristics that are important to the particular project.

Typical characteristics
■ Quality and availability of the energy resource;
■ Fuel cost;
■ Capital cost, including balance of system components such as
batteries, inverters, conveyance systems, towers, and fuel storage;
■ Lifespan of major components;
■ Equipment availability, including replacement parts;
■ Maintenance requirements;
■ Human capital requirements—including technical and business
skills;
■ Environmental and land-use requirements;
■ Design difficulty.

Chapter 12Design and Implementation of Off-Grid Systems

Design Approaches
■ Our primary focus is the electrical design
■ There are two basic approaches to designing a mini-grid:
– intuitive
– numerical
■ The design of a mini-grid is iterative
■ Several designs are considered before a final design is selected
■ Each approach requires an estimation of the load and energy
resources as described in the previous chapter

Intuitive Approach
■ Intuitive design approaches have low input data requirements and allow the
mini- grid to be designed using relatively simple calculations
■ The calculations can be based on “rules of thumb” or formalized standards,
such as IEEE 1526, IEEE 1013, and IEC TS 62257
■ Intuitive approaches, however, provide limited insight and feedback as to
how design choices affect the performance of the system, particularly its
reliability
– The final design may have lower reliability than needed, or higher
reliability than necessary, wasting resources.
■ A complete design of the mini- grid would include specification of
– wire sizes
– switches
– protection equipment and grounding
– physical layout
– mechanical and civil design
– the distribution system.

An Example of Intuitive Approach
1 Project Overview
■ Assume that after screening and site assessment, the community of
Mwase in the southern region of Zambia has been selected for a mini-
grid implementation
■ Mwase is a fictional community. Its characteristics, however, are drawn
from real off-grid communities.
■ The population of Mwase is approximately 3000 people
■ The nearest electrified town, Bona, is 23 km away
■ The road to Mwase is not well maintained and becomes impassable
during the rainy season (roughly November to February)
■ The people of Mwase primarily support themselves by fishing, raising
livestock, and farming. It is anticipated that in addition to serving
households, the mini-grid will supply a small number of businesses
and a school located on the far side of town. One entrepreneur will
start a barber shop/hair salon which will use a high-power blow dryer;
another will sell refrigerated drinks from their shop.

1 Project Overview
■ The households have indicated they will primarily use the electricity for
lighting and watching television in the evening.
■ These users are located along the main pathway through the town.
■ The households to be connected to the mini-grid are arranged in two
dense clusters along the pathway.
■ No need of a three-phase connection supplied with a single-phase 230
V, 50 Hz connection (matches that of the national grid) the appliances
readily available in the country can be used
■ A secure powerhouse will be constructed.
■ The energy resource estimation
– WECS, MHP, and biomass are not technically viable
– a gen set or a PV array are possible sources
■ The preliminary design: use the PV array system (assuming that the
logistics of maintaining the fuel supply for a gen set are not practical)

2 Load Characterization

3 Architecture Selection
■ The only generation source is the PV array, and so the system is DC-
coupled
■ The users require AC service need an inverter
■ A battery bank will supply the load during evening hours
■ Because the system will be larger than a few hundred watts, a charge
controller with MPPT will be included
■ The PV array will be installed on top of the power house, and the
battery bank and all the DC components will be securely enclosed
within.

4 Voltage Level Selection
■ The AC bus voltage is already determined: 230 V at 50 Hz, single-
phase
■ The DC bus voltage is set by the battery bank. Here we consider
several factors:
– Nominal voltage of selected battery chemistry (see Chap. 8)
– Electrical distance between DC bus components
– Power capacity of the system
– Safety
– Compatibility with DC bus components
■ We will assume that the only large-capacity batteries easily available in
the country are lead–acid. Therefore, the nominal voltage level will be
an integer multiple of 2 V. The battery bank is in a secure area, and so
both flooded and valve-regulated lead–acid are viable options. We will
use AGM valve-regulated batteries because of the low maintenance
requirements.

4 Voltage Level Selection
■ The battery bank will be located in the power house near the PV array and
inverter, and so we do not need to be concerned about excessive voltage
drop or losses
■ The capacity of the PV array will be several kilowatts. To keep the wires from
being too large, a higher voltage is preferred
■ To be compatible with most charge controllers and inverters, we should
select either 12, 24, or 48 V
■ A common rule of thumb is
– 12 V for systems whose load is less than 1 kWh/day
– 24 V for load between 1 and 4 kWh/day
– 48V when the load exceeds for 4 kWh/day
– Exceptionally large systems might use 96 V or above
■ We therefore select a nominal 48 V DC bus voltage
■ With the DC bus voltage set, the current ratings of the charge controller and
inverter can be calculated.

5 Inverter Selection
■ Recall from an inverter facilitates the flow of power from the DC bus to the AC
bus. Inverter parameters that require specification by the designer are shown in
Table 12.3
■ In general, the sum of the power ratings of the inverters and generators
connected to the AC bus must be at least as large as the peak load. The
minimum power rating of the inverter serving Mwase must therefore be 1.94 kW

■ However, this value requires adjustment. We will apply a design margin to
mitigate the possible underestimation of the peak load
■ The inverter will not be in a temperature-controlled environment. Most
inverters are rated based on a certain ambient temperature, usually 25◦C. If it
is expected that the ambient temperature will consistently be higher than the
rated temperature, then an inverter with a larger capacity is needed
■ We will select a high design margin of 0.20 to also account for the effect of
temperature:
Inverter Power Requirement = Peak Load × (1 + Design Margin)
The minimum required rating becomes 1.94 × (1 + 0.20) =2.33 kW
Assume a power factor of 0.85 so that the inverter rating in voltamps is
2.33/0.85 = 3.74 kVA

■ We next compute the maximum input current on the DC side of the
inverter:
■ Inspection of Mwase’s estimated load profile in Fig. 12.3 shows that
for most hours the load is between 400 and 600 W. This is roughly 20
to 25% of the required inverter rating. Given the general shape of
inverter efficiency curves shown in Sect.9.9.5, we will assume the
average efficiency of the inverter is 85%. This estimate could be
improved by calculating the efficiency for each hour of the load profile
using the selected inverter’s efficiency curve and averaging the result.
■ From (12.2), the maximum current on the DC side of the inverter is
For convenience, we will round the maximum current up to 58 A.

■ There are a few other inverter parameters that we should specify
■ We are not sure if the appliances will tolerate high levels of distortion,
so a modified sinewave inverter should not be used
■ We will only consider inverters whose total harmonic distortion (THD)
does not exceed 5%
■ The users only need single-phase service, we will use a single-phase
inverter
■ The mini-grid’s architecture is such that power will never flow from the
AC bus to the DC bus, so the inverter need not be bi-directional
■ The battery bank needs protection from deep discharge the inverter
must have a low-voltage disconnect (LVD) feature

■ Most inverters have a “peak power” or “surge power” rating in addition to their
continuous rating. This refers to the power that the inverter can supply for a
short duration (often a few seconds to several minutes).
■ Depending on the nature of the load, especially if the peak load is due to motor
starting, the peak rating can be used instead of the continuous rating
■ For Mwase we will use the continuous rating, which is a more conservative
approach
■ We also note the no-load power of the inverter is 25 W. Over the course of the
day, the inverter will consume 600 Wh, which represents about 6% of the daily
load. The load estimate can be increased to account for this in later iterations
of the design. In the case that a single inverter is not capable of supplying the
required peak power, multiple inverters can be connected in parallel. The
inverters should be from the same manufacturer, and they should be capable
of synchronizing with each other. If this is not possible, then separate AC buses
can be created, with each inverter supplying power to its own AC bus and the
subset of users connected to it.

Battery Bank Design
■ The battery bank is designed independently of the PV array. The
following steps can generally be applied regardless of the energy
conversion technology selected.
■ The three most important factors affecting the design of the battery
bank are
(1) the nominal voltage of the DC bus (specified already)
(2) the discharge current (determined from the inverter input current)
(3) the required reliability (“Days of Autonomy” )
The Days of Autonomy: the number of days that the battery bank can
supply the average load before being depleted, assuming it is not
recharged during this period. This is an extreme scenario. It could occur if
the PV array is damaged and the repair was delayed.

Battery Bank Design
■ A system designed using more Days of Autonomy will have higher
reliability than one with fewer. Increasing the Days of Autonomy
increases the required battery bank capacity. However, the reliability
does not necessarily proportionately scale with the Days of Autonomy.
In other words, doubling the Days of Autonomy does not necessarily
double the reliability. The specified Days of Autonomy is typically
between 2 and 12. We will select just two Days of Autonomy to reduce
costs.
■ It is convenient to specify the battery bank capacity requirement in
terms of amphours. The average daily load required from the battery
bank, in terms of amphours, is
■ For Mwase, this is

Battery Bank Design
■ At a minimum, the battery bank must be capable of supplying the
average battery load for the specified Days of Autonomy. This must be
the case even at the end of the considered 5-year period
■ However, the battery bank’s maximum capacity will naturally decrease
over time. Most manufacturers define the maximum capacity at the
end of life to be 80% of the initial capacity. The minimum-rated
capacity of the battery bank is therefore
■ where x is the discharge current the capacity is associated with, which
will be discussed shortly.
■ The battery bank for Mwase must have a minimum capacity of

Battery Bank Design - Configuration
■ The number of batteries in series is computed from
■ The given battery has a nominal voltage of 6 V. Therefore, each string
will have 48/6 = 8 series-connected batteries.
■ The number of strings required is found from
■ Using a capacity of 200 Ah, 827.21/200 = 4.13 strings are needed,
which we round up to 5 strings. With this design, each string supplies
58/5 = 11.6 A during the peak load.
■ The battery bank therefore has 5 strings of 8 batteries, for a total of 40
batteries.

Battery Bank Design
■ We note that it would also be reasonable to use four strings instead of
five
■ Recall that we conservatively used the maximum discharge current to
determine the discharge rate
■ If we used the average discharge current instead, we would find that
four strings would be viable. This would reduce the material cost of the
battery bank cost by 20%. It also results in fewer batteries

7 Energy Source Design
■ The guiding principle of energy source design is that the source should
be capable of supplying enough energy to supply the expected average
load accounting for generation and storage losses
■ The system is designed around the month with the lowest capacity
factor. If the load has a seasonal component, then designs are
produced for each month or season, and the design with the largest
energy source rating is used.
■ The average daily insolation by month for Mwase was found by
consulting a solar database. The data are provided in Table 12.7 for
three different array tilts.

7 Energy Source Design - Tilt Selection
■ The tilt of PV array is selected by comparing the insolation data in
Table 12.7
■ The month with the lowest insolation for each tilt is highlighted in bold.
■ We select the tilt that corresponds to the largest of these three values.
That is, we select the tilt with the greatest minimum monthly
insolation. In this case it corresponds to an array tilted at the latitude,
whose minimum insolation occurs in January with 5.08 kWh/m2
/day.

7 Energy Source Design - Capacity Factor
■ The capacity factor is related to the insolation as
■ if all losses are ignored, the PV array must be rated at no less than
■ For Mwase, this corresponds to
■ This represents an idealized case
■ A higher rating is needed to account for losses

7 Energy Source Design - Generation and Storage Losses
■ The actual usable energy production from a PV array will be less than
EPV. This is due to a number of factors:
– Array shading, including dust
– Wire and connection resistive losses
– Parasitic losses (stand-by consumption of controllers, monitors,
data acquisition systems, and other devices)
– Module mismatch (caused by PV strings or modules having
different maximum power points)
– Array degradation over time (aging)
■ The production is also affected by coincidence of the load and the
irradiance. Recall that during the absorption charging stage, the power
to the battery bank is intentionally limited. If the load is also low during
this time, the PV array production will be reduced (throttled). Further,
any energy stored incurs losses associated with the battery

7 Energy Source Design - Generation and Storage Losses
■ It is difficult to estimate the generation and storage losses. Typical ranges are
shown in Table 12.8. Whether it is prudent to select a value toward the low end
■ Estimates are made for each type of loss, and the results are summed
to determine the total loss KL. It is obvious that 0 ≤ KL ≤ 100.
Table 12.8
Typical
generation and
storage losses

7 Energy Source Design - Temperature Effects
■ The power output of a PV module decreases with temperature. The
temperature of a PV cell during the daytime is usually much higher
than the ambient air temperature
■ A reasonable range of operating temperatures is between 30◦C and
60◦C. Given a typical temperature power coefficient αp of −0.5 % per
degree Celsius (or Kelvin) above 25◦C, the presumed power output of
the PV array should be reduced by 2.5% to 17.5%
■ A value within this range can be selected based on an informal
assessment of the climate (warmer locations warrant more severe
reduction). However, more accurate estimation can be made using the
specification of the PV array and temperature data
■ The required rating becomes

7 Energy Source Design - Design Margin
■ The reliability of the mini-grid can be improved by sizing the PV array such that
its energy production capability somewhat exceeds the energy supplied by the
battery bank to the load each day. The PV array will be able to supply additional
energy to the battery bank in case it is deeply discharged or if the insolation is
below average for several consecutive days.
■ The PV array design margin, KPV, generally ranges from 0.1 to 0.2 (10 to 20%)
for systems with non-critical loads or with consistent average daily load and
consistent insolation. Systems with higher reliability requirements or with
inconsistent load or insolation should use higher values, perhaps up to 0.4.
■ The PV array requirement after accounting for the design margin is

7 Energy Source Design – Summary

8 Charge Controller Selection
■ The major parameters that must be specified are in Table 12.11 in last
slide
■ The charger satisfies the voltage and current constraints, but not the
power constraint. We can either select a larger charge controller or use
two in parallel. We will consider two in parallel. One controller will have
three strings and the other two.
■ This also reduces the maximum short-circuit current that each charge
controller could be exposed to.
■ We next check that the current supplied by the charge controller does
not exceed the maximum current recommended by the battery’s
manufacturer. Per the battery specifications, the maximum charging
current is 20% of the C20 rating, which equals 44 A. The battery bank
has five strings, and so the maximum battery bank charging current is
44 × 5 = 220 A. The maximum charging current per charge controller is
60 A (120 A total), and so we are not concerned with exceeding the
battery bank charging current limit.

8 Charge Controller Selection
■ There are few other parameters that should be specified.
■ Due to the size of the PV array, the additional cost of using a charge
controller with MPPT functionality is justified.
■ To prolong the lifespan of the battery bank, we should select a
charge controller that uses a three-stage charging algorithm. The
charge controller’s absorption and float voltage set-points are set
according to the battery manufacturer’s specifications.
■ This completes one iteration of an intuitive design process. This
design serves as a starting point for economic and other analyses.
Based on the results of these analyses, new designs will be
produced using different inputs—for example, serving additional or
fewer users—or changing assumptions and specifications such as
the Days of Autonomy.

9 Cost Estimate
■ The material costs will vary widely from one country to the next and
from one vendor to the next. Table 12.12 gives an indication of
costs that might be considered typical in Zambia in 2016, inclusive
of the value-added tax
■ Additional costs are shown Table 12.13. Balance of system (BoS)
components typically include power house wiring, outlets, switches,
circuit breakers, grounding rods, and surge arrestors.
■ The total estimated cost is US$19, 335 + US$7290 = US$26,625.
This excludes the cost of the distribution system and any costs
associated with the site assessment and engineering of the
system.
■ If this cost is too high, then alterations to the design can be made:
– reduce the number of battery bank or PV module strings
– designed to serve fewer users or to supply electricity at a
lower access tier

Numerical Approach
■ Numerical approaches can be applied to simple systems, but they
are particularly useful in designing hybrid systems with complex
control
■ The numerical design approach relies on a computer program to
guide the design of the mini-grid. Most are simulation-based. That is,
they simulate the operation of a mini-grid over a period of time and
provide the user with a summary of the results
■ The programs do not design the mini-grid. They are better thought of
as aides or tools assisting the designer
■ The programs require engineering judgment at both ends of the
process—defining the technical and economic environment in which
the mini-grid will operate and interpreting the results provided by the
program.
■ Numerical approaches allow the designer to make better-informed
decisions than intuitive approaches. The programs can provide the
designer with information regarding the reliability and cost and other
technical and economic information

Numerical Approach
■ However, the programs have heavier input data requirements than
intuitive approaches
– Require the hourly or daily variability of the load to be specified
– If the input data provided by the user has a wide range of
uncertainty, as it often does, then so does the output of the
simulation
■ There is no guarantee that the actual mini-grid will perform as
simulated, especially if the load or resource is substantially higher or
lower than input into the program
■ Despite this caveat, computer-aided simulation can be very beneficial
when the user understands the limitations and capabilities of the
program
■ Uncertainty in input data can be guarded against by simulating many
load and resource scenarios, for example, high, medium, and low
projections
■ The design that best balances cost and risk is selected
■ Computer-aided design programs are continually improving and will
likely play an increasing role in mini-grid design in the future.

Numerical Approach
■ There are a few commercial-grade computer programs that can
assist in mini-grid design
– The most popular are HOMER and RetScreen
– There are at least a dozen other programs developed by
universities, but these tend not to be well-supported.
■ Most computer-aided tools use a time-based simulation of the mini-
grid. This means that the program simulates the operation of the
mini-grid at discrete time steps over a period of time
– The simulation period is usually 1 year but can be longer or
shorter
– Most programs use a time step of 1 h
– A power flow model, like that developed in the previous chapter,
is used to simulate the operation and control of the mini-grid at
each time step
– The different programs might use slightly different models, but
the general approach is similar

Thanks, this course ends here.

Chapter 12Design and Implementation of Off-Grid Systems

More Related Content

Similar to Chapter 12Design and Implementation of Off-Grid Systems (20)

Recently uploaded (20)