Load Forecasting: Introduction, Classification of Load, Load Growth Characteristics, Peak Load Forecasting, Extrapolation and Co-Relation methods

of load Forecasting, Energy Forecasting, Reactive Load Forecasting, and Impact of weather and Factors affecting load Forecasting, Annual, Monthly and Total Forecasting.

LOAD FORECASTING

1. Introduction

Role of Electrical Energy To maintain a highly developed

NEEDFOR LOAD FORECASTING

industrialized country. Growth rate in generation drop off as leveling population growth and energy consumption per capita. Utility industry reaching in a no-growth state . Utility industry constantly scrutinizes the dynamic growth pattern. Determine the individual load characteristic. Plans expansion of power systemstarts with Forecasting

FORECAST OF ANTICIPATED FUTURE REQUIREMENTS

A. DEMAND FORECAST
To determine:capacity of a. Generation b. Transmission & c. Distribution systems. To establish:Procurement policies for construction capital

B. ENERGY FORECAST
To determine:a. Type of facilities required b. Future fuel requirement

DEFINITION
LOAD :-

LOAD DEMAND FORECAST

A general term meaning either Demand or Energy.

DEMAND:-

Time rate of change of Energy.

FORECAST :-

Projected load requirements. (defines future loads- to permit important system expansion to be made)

FORECAST ACCURACY
Crucial to any Electric utility. Dictates :

Timing & Characteristics of Major System Additions. Too Low Forecast - Low Revenue (from sales to neighboring utilities or in load curtailment) Too High Forecast Severe Financial problems (due to excessive investment in the plant not fully utilized i.e. low capacity factor av. Energy supplied/max. Energy capability) Accurate Forecast depends on judgment of Forecaster. Impossible to rely strictly on analytical procedure to get accurate forecast.

2. CLASSIFICATION OF LOADS

Total Forecast:

Combined Forecasts for Various Classes of Costumers(Loads)

Classification ( Broad):-

a. Residential b. Commercial c. Industrial d. Others

ENERGY USE

a. For Domestic Purpose

By a. Residential b. Commercial c. Industrial d. Others

b . For Commercial Purpose

c. For Industrial Purpose

d. Municipalities or Divisions of State

& Federal govt. for street and Highway lighting; Sales to Public authorities and to Railroads & Railways, Sales for resale, & Interdepartmental sales.

Subdivisions

Rural
Residential Customer

Urban Others

NEED FOR CUSTOMER (LOAD) CLASSIFICATION FOR FORCASTING PURPOSE

Classifications Overlap.

i.e, customers in the same class do not have characteristics unique to that class. i.e, Classifications are not mutually exclusive. Within the broad class subdivisions may be defined.

CUSTOMER CLASSIFICATION FOR FORECASTING PURPOSE

Type of use Level of use Rate schedule or

Rate schedule :

. Potentially a viable way

-Lump similar types of customers into same rate category

Geographic area

3.Load Growth Characteristics

Residential
Most constant annual growth rate Most seasonal fluctuations

Commercial
Characterized by seasonal fluctuations a. Residential b. Commercial c. Industrial i.e. extensive use of air conditioning & space heating

Industrial
Base Loads No weather dependent variations

Growth trend almost stable.

4.

Load Growth Characteristics other effects

Residential & Commercial

Other seasonal variations
Caused by - Economical - Demographic effects

Industrial
-Variations in load requirement

5. Peak Load Forecasting

Can be done by:
a. Using Forecasted Energy & Load Factors b. By combining Forecasts of appropriate load components or from historical total load data. c. Using average or extreme weather conditions . d. Using simple methods & mathematical procedures

a.

Using Forecasted Energy & Load Factors

Two Methods:i. Forecast Energy & Obtain Demand Forecast from it. ii. Forecast Peak Demand directly

b. By combining Forecasts of appropriate load components or from historical total load data. Combining Forecasts
Types of

Historical total load data

Easier to use More indicative of overall

customers , geographic areas Abnormal conditions in growth trends of individual component can be detected. Can prevent misleading forecast conclusions.

growth trends

c. Using average or extreme weather conditions

Average weather conditions Extreme weather conditions
Determine extreme peaks:

Average weather

using historical differences of nominal & extreme weather conditions.

d.

FORECASTING METHODOLOGY
quantitatively defining future loads .

A simple systematic procedure for

Techniques ( based on time period ):

1. short term 2. inter-mediate & 3. long-term technique.

*Forecast implies intermediate-range forecast.

FORCASTING TECHNIQUES
THREE BROAD CLASSES

BASED ON:
1. EXTRAPOLATION . 2. CORELATION 3. COMBINATION OF BOTH

1. Using historical Data to determine what will happen in the future 2. Relate system loads to various demographic & economic factors

EXTRAPOLATION
the process of constructing new data points outside a

discrete set of known data points. similar to the process of interpolation, which constructs new points between known points. the results of extrapolations are often less meaningful, and are subject to greater uncertainty.

EXAMPLE
Extrapolation may also apply to human experience

to project, extend, or expand known experience into an area not known or previously experienced so as to arrive at a knowledge of the unknown
e.g. a driver extrapolates road conditions beyond his

sight while driving.

FURTHER CLASSIFICATION
deterministic extrapolation - no attempt is made to account for

1. DETERMINISTIC

random errors in the data or in the analytical model

Probabilistic extrapolation

2. PROBABILISTIC OR STOCHASTIC

To quantify uncertainty of extrapolated results: Statistical entities: mean & variance Basic technique PROBABILISTIC EXTRAPOLATION

EXTRAPOLATION TECHNIQUE
Fitting Trend curves:

- to Basic Historical Data adjusted - to reflect the Growth Trend itself. To obtain the forecast: - Evaluate the Trend Curve Function - at the Desired Future Point

Example illustration of the extrapolation problem, consisting of assigning a meaningful value at the blue box, at x = 7, given the red data points.

STANDARD ANALYTICAL FUNCTIONS

To Fit Trend Curve:

1. Straight Line 2. Parabola 3. s Curve 4. Exponential 5. Gompertz

Y = a + bx Y = a + bx + cx Y = a + bx + cx + dx Y = cedx Y = In-1(a + cedx )

To find coefficients & exponents of a forecast function: method of Least Squares-curve fitting technique.

REASONS FOR UNCERTAINTY

TWO SOURCES :

UNCERTAINTY in a. historical data b. the analytical model chosen to describe the growth in load SOLUTION: To forecast the Trend obtain: the best estimate of the model describing the trend with Regression Analysis

Extrapolation methods
Choice :

which extrapolation method to apply relies on a prior knowledge of the process that created the existing data points.
Methods :

Linear Extrapolation Polynomial Extrapolation Conic Extrapolation French curve Extrapolation

ENERGY FORECASTING
USING:
CORRELATION & EXTRAPOLATION TEMPERED WITH SOUND PROJECTIONS OF FUTURE CONDITIONS.

TOTAL ENERGY FORECAST:

FORECAST FOR THREE MAJOR CLASSES OF CUSTOMERS:
* RESIDENTIAL * COMMERCIAL * INDUSTRIAL

RESIDENTIAL SALES FORECASTS

MAJOR FACTORS THAT AFFECTS RESIDENTIAL ENERGY

REQUIREMENT:

1. RESIDENTIAL CUSTOMERS 2. POPULATION PER CUSTOMER 3. PER CAPITA ENERGY CONSUMPTION

FORECAST OF RESIDENTIAL SALES = EACH FACTOR PER WEEK,MONTH etc. * EACH FACTOR.

HOW TO OBTAIN THE VALUES OF THE THREE FACTORS

BY

a. simple curve fitting method b. regression analysis (methods a & b referred as POPULATION method) c. synthetic method ( requires a more detailed look at each customer) ( major factors: 1. saturation level of major appliances 2. average energy consumption per appliance 3. residential customers )

Regression analysis: Goal & Example

Goal : to determine the values of parameters for a function

that cause the function to best fit a set of data observations that you provide. In linear regression, the function is a linear (straight-line) equation. Example: assume the value of an automobile decreases by a constant amount each year after its purchase, and for each mile it is driven the linear function that predict its value is value = price + depage*age + depmiles*miles
(the dependent variable on the left side of the equal sign) as a

function of the two independent variables which are age and miles:

..Example
a value of 16000 for price, -1000 for depage, and -0.15 for

depmiles, then the function value = 16000 - 1000*age - 0.15*miles estimate the value of a car with a known age and number of miles. "deviation'' or "residual:the difference between the actual value of the dependent variable and its predicted value for a particular observation (error of the estimate ) "least squares'' regression fit: determine the values of the parameters that minimize the sum of the squared residual values for the set of observations

plot of a linear function fitted to a set of data values. The actual data points are marked with ''x''. The red line between a point and the fitted line represents the residual for the observation.

COMMERCIAL SALES FORECASTS

Commercial establishments: service oriented

- closely related to growth pattern of residential sales Methods: 1. Extrapolate ratio of commercial to residential sales & multiply by residential sales forecasts. 2. Extrapolate historical commercial sales.

INDUSTRIAL SALES FORECASTS

The most difficult Reasons:

- Tied very closely to the overall economy - Over selected period economy unpredictable

METHODS TO DEVELOP INDUSTRIAL SALES FORECASTS

1. Multiply forecasted production levels

by forecasted energy consumptions per unit of production.

2. Multiply forecasted industrial workers

by forecasted energy consumptions per worker.

PEAK DEMAND FORCASTING

Extrapolating Historical demand data. Weekly peak demand forecast & to get monthly or annual forecast
Steps to consider

1. determine seasonal weather load model 2. separate weather-sensitive & non- weathersensitive components of weekly peak demand using weather load model 3. forecast mean & variance of non- weathersensitive component of demand

Steps to consider.
4. extrapolate weather load model & forecast mean & variance of weather-sensitive component 5. determine mean, variance & density function of total weekly forecast 6. calculate density function of monthly or annual forecast