Power System planning and

Management
Lecture 3 :
Load Forecasting

Yoseph Mekonnen (Ph.D.)

Page 1
Outlines
Introduction
Load Characteristics
Load Driving Parameters
Types of Load Forecast
Load Forecasting Methods

Page 2
 Introduction
The purpose of power generation is to supply as much load
as possible.
Both the energy (MWh, kWh) and the power (MW, kW) are
the two basic parameters of a load and by load, we mean the
power.
If the load shape is known, the energy can be calculated
from its integral.

Forecasting refers to the prediction of the load behavior

for the future.
Power system design starts with load forecasting
Demand forecasts are used to determine the capacity of
generation and transmission.

Page 3
 Load Characteristics
The Load Characteristics is divided into different
categories.
The Most Common ones are:
Daily Load Variation (Hour-by Hour)
Weekly Load Variation(Day-by-Day)
Seasonal Load Variation (Summer, Winter..)
Yearly Load Variation (2015, 2016,…)

Page 4
 …Contd..
Daily Load Variation
The daily load curve of a transmission substation has a
general shape similar to Fig.

The daily load variation has an unstable and non linear shape
than the weekly load variation.
It is evident that the load shape of a working day is
significantly different from that of a weekend day.
Moreover load shape of working days may be different due
to, the weather conditions
Page 5
 …Contd..
Seasonal Load Variation
The seasonal load curve of a transmission substation has a
general shape similar to Fig.

The seasonal load variation only concentrates on the peak

values of the hour by hour load variation.
Hence there is uncertainties associated with seasonal
forecast

Page 6
 Load Uncertainty
The weather variation is one of the sources of load
variability.
In a cloudy day, the use of cooling devices like air
conditioner and ventilators may not be necessary if the
temperature is not severe.
At such particular moment and time, the load demand may
fall to its lowest point.
In contrary to this, during a sunny day end users switch
on both cooling machines and their refrigerators that incur
more power in a residential area .
When the weather is cloudy, generation from solar plant
decreases and cannot supply some of the loads, which
results the variability in load.

Page 7
 …Contd..
When there is a high outage in a given system, customers
switch on every chargeable device immediately after system
restoration in order to use later.
Such kind of act boosts the load demand and makes the
generation system to not fully support load in addition to
the presence of variable generation systems.
The cost of electricity is also another source for load
uncertainty.
Some power companies provide low off peak rate, which
forces customers to shift the peak load to off peak load
duration.
A highly deficient generation and transmission network
hides the effect of load uncertainty coming from forecast
error.
Page 8
 …Contd..
Load forecast uncertainty consideration has to
incorporate the variation of the power demand at the
individual load point in a bulk electric system.
As a result, load variation must be considered as a
variable input during load flow analysis of a bulk power
system network.
If for planning purposes, we are going to predict the load
variations for the next several years (say 10 years or more),
we do not bother the daily variations.
Instead we may have to predict, say, the summer and the
winter peaks, of the coming years.
 It means that for a 10-year prediction, only 20 data are
required.

Page 9
 Load Driving Parameters
There are Parameters affecting the forecasted load of
future.
These driving parameters are quite a few.
Some typical ones are as follows:
Time factors such as:
•Hours of the day (day or night)
•Day of the week (week day or weekend)
•Time of the year (season)
Weather conditions (temperature and humidity)
Class of customers (residential, commercial, industrial,
agricultural, public, etc.)
Special events (TV programmes, public holidays, etc.)
Population
Page 10
 …Contd..
Economic indicators (per capita income, Gross National
Product (GNP), Gross Domestic Product (GDP), etc.)
Trends in using new technologies
Electricity price.
If the electricity price is predicted to be high, it results in
a reduced forecasted load.
If the economic indicators such as GNP and GDP show a
promising future and new electricity based
appliances/technologies are appearing in the market, the
electricity consumption may increase nearly in all class of
customers

Page 11
 …Contd..
Load Driving parameter

Page 12
 Classification of Load Forecast
Normally load forecasting is classified into three different
categories.
Short Term Load Forecast
Mid Term Load Forecast
Long Term Load Forecast

Page 13
 …Contd..
Short Term Load Forecast (STLF)

The STLF methods are used for hour-by-hour predictions.

STLF can be used for 1 day to 1 week.
Some driving parameters may be ineffective or ignored for
STLF.
GDP may have strong effects on LTLF; while ineffective in STLF
TV programmes are effective in STLF but ineffective in LTLF
TLF normally results in hour-by-hour forecast (for 1 day to 1
week).
The price forecast for STLF is more accurate than that of
MTLF. The same is true for weather forecast
Example: Peak Load and off Peak Load of the Day.

Page 14
 …Contd..
Mid Term Load Forecast (STLF)
MTLF normally results in daily forecast (for several weeks
to several months).
Normally the peak of the day is forecasted for several
weeks.
Long Term Load Forecast (LTLF)
The LTLF used for several years.
LTLF focuses on monthly or seasonal forecasts (the peak of
the month or the season) for several years from now.
GDP have strong effects on LTLF.
Due to inaccuracies involved in long-term driving parameters,
it is of common practice to perform LTLF for several
scenarios (such as various GDPs, weather forecasts, etc.).
Page 15
 …Contd..
Factors Affecting Load Forecast
Historical Data
City Plan
Population Growth
Community Growth
Industrial Plan
Load Density

Page 16
 …Contd..
For starting Load planning thee following data should be
collected:
1. A detailed map of the area
2. No of houses, population and new construction.
3 Details of hospitals transportations railway station
4. The expected no of shops, post offices,….
5. Types and nos. of industries,…
6. The cultivated area and irrigation facilities,..
7 Development Programs implemented and to be
implemented

Page 17
 …Contd..
Spatial Load Fore Casting
It is the the geographical distribution of loads.
Planning for the future expansion of a power system
involves determining both the capacities and the locations
of future components; namely, generation facilities,
transmission/sub-transmission/distribution lines and/or
cables and various substations.
This requires forecasting the future loads with geographic
details (locations and magnitudes).
Spatial load forecasting is accomplished by dividing utility
system into a number of small areas and forecasting the
load of each.
Example: substations, feeders and/or grid location.

Page 18
 …Contd..
Once the load of each cell is predicted, the electric load
of the system (or a larger geographical area) can be
predicted.

Page 19
 …Contd..
Load Estimate of a Colony

Page 20
 Load Forecasting Methods
The main focus is on Long Term Load Forecast.
The LTLF methods are basically:
Trend Analysis
Econometric Modeling
End-use Analysis
Combined Analysis.

Page 21
 …Contd..
Trend Analysis
The trend extrapolation method uses the information of
the past to forecast the load of the future.
A curve fitting approach may be employed to find the load
of the target year.
This approach is simple to understand and inexpensive to
implement.
However, it implicitly assumes that the trends in various
load driving parameters remain unchanged during the study
period.
For instance, if there is a substantial change in economic
growth, the approach fails to forecast the future load,
accurately.

Page 22
 …Contd..
More weights may be attached to the loads towards the
end of the past period.
In this way, the prediction may be improved.

Page 23
 …Contd..
Econometric Modeling
In this approach, initially the relationship between the load
and the driving parameters is estimated.
The relationship may be nonlinear, linear; additive or in the
form of multiplication.
This relationship is established based on available
historical data.
Various driving parameters may be checked to find the
ones that have the dominant effects.
A typical nonlinear estimation is:

Di  a( Per Capitaincome)bi ( Population)ic ( Electricity Pr ice)id

Page 24
 …Contd..
where i denotes the year and a, b, c and d are the parameters
to be determined from the historical data.
Once this relationship is established, the future values of
the driving variables (i.e. per capita income, population,
electricity price, etc.) should be projected.
Di for a future year can then be determined.
This approach is widely used and may be applied to various
customer classes (residential, commercial, etc.) and to the
system as a whole.
It is relatively simple to apply.
The drawback is the assumption of holding the relationship
established for the past to be applicable for the future.
In this way, the influence of any new driving parameter
cannot be taken into account.
Page 25
 …Contd..
End-use Analysis
This type of analysis is mostly confined to residential loads
but may be applied with some modifications to other load
classes, too.
As a simple example, if refrigerator is concerned, based on
the number of households and estimating the percent of
households having a refrigerator, the number of
refrigerators for a future year may be estimated.
Following that and based on average energy use of such an
appliance, the total energy consumption of refrigerators
may be estimated.
This approach may lead to accurate results if its extensive
accurate data requirements can be provided.

Page 26
 …Contd..
Load Forecasting for a Regional Utility
Assume a given area consists of eight sub regions (area).
Each area consists of some subareas, supplied through
some substations, either existing or new.

Page 27
 …Contd..
The aim is to predict the peak load, as well as, the energy
demand of the regional utility for 10 years from the
current year; with a time step of 1 year.
The process starts from the subareas; moving upwards to
reach the load for the region.

It is assumed that the geographical characteristics of the

subareas as well as their load data for the last 10 years are
known.
The first step is to identify what type of load is contained
in the sub areas.

Page 28
 …Contd..
Subarea consists of the following three types of loads:
Urban
Rural
Large customers

Urban load Consists of:

Residential
Commercial
Public
Small industrial
Distribution losses

Page 29
 …Contd..
The rural types of loads, mainly consist of
 Residential
 Agricultural
 Others (small industrial, public, etc.)
The residential part may be estimated based on the
estimated number of homes and the estimated power
consumption of each home

Page 30
 …Contd..
Historical data as well as extensive data from the regional
departments, in charge of the above mentioned sections,
are required to reach at reasonable predictions.
An urban load is not, actually, concentrated at a specific
geographical point and is distributed throughout the urban
territory. These points have to be observed.

Page 31
 …Contd..
Large customers load
They are considered separately, as they do not obey any
specific rule, in terms of, the forecasted loads.
They may be either existing or new.
The future loads of existing types may be estimated based
on their previous and foreseen performances.
The loads of new types are determined based on the
demands of their respective contracts with the utility.
Both require extensive data gathering and communications
with the large customers and the departments in charge of
large customers.
They are typically the customers with more than 1 MW
demand.
Page 32
 …Contd..
Numerical Data for Area E
This area consists of four subareas.
There are two main urban loads in subareas 1 and 2 (UE1
and UE2). The Load of the area varies as shown below.

Page 33
 …Contd..
The full urban load is shown below for some selected years.

Coincident demand is the energy demand required by a given

customer or class of customers during a particular time
period

Page 34
 …Contd..
The full agricultural load is shown below for some selected years.

Page 35
 …Contd..
Rural Load

Page 36
 …Contd..
Results for area E

Page 37
 …Contd..
Results for area E (Sub area based)

Page 38
 …Contd..
Regional

Page 39
 …Contd..
Regional (Agricultural)

Page 40
 …Contd..
Regional (Overall)

Page 41
Thank You!

Page 42