MODULE II

LinssT

Alex

Syllabus

Problems facing electricity industryThe very common electrical problems are:

• Intermittent power

• Power surges and spikes

• Sags, dips and outages

• Redundant wiring

• Overloaded circuits

Intermittent powerIntermittent power is a symptom of a wiring problem. The cause of this common electrical problem is loose wiring. To check for this type of problem, trace the electrical cable from the unit to the plug. Look for any sign of wear, fray, or exposed wiring.Take apart the electrical item to locate the connection point of the power to the unit

Power surges and spikes

Power surge -An oversupply of voltage from the powercompany that can last up to 50 microseconds.Although surges are very short in duration, they oftenreach 6,000 volts and 3,000 amps when they arrive at theequipment

Power Spike- An instantaneous power increase ofgreat magnitude, such as that cause by a lightning strike.A spike can arrive via power lines, phonelines, network lines, or the like, and may causeextensive damage if protective measures are not taken.

Plug-In Surge Protectors are used as a solution againstthese problems.

Plug-In Surge Protectors shield individual pieces ofelectronics in your home from spikes, surges and electricalnoise that originate inside the home, and offer measure ofprotection from outside disturbances.

Sags, dips and outages

A voltage sag or voltage dip is a short duration reductionin rms voltage which can be caused by a short circuit,overload or starting of electric motors.

A voltage sag happens when the rms voltage decreasesbetween 10 and 90 percent of nominal voltage for one-half cycle to one minute.

Outages (or blackouts) are periods when there is noelectric power. These can last from less than a second tominutes or even longer.

While sags and dips can be caused when an electricaldevice draws power as it is turned on, outages are usuallycaused by severe weather, accidental damage to yourpower company equipment or electrical short circuits.

To protect equipment against these disturbances, as well asmost other types, Uninterruptible Power Supply (UPS)Units can be useful in saving both equipment and work.

Redundant wiring

Redundant wiring is a very common electrical problemthat occurs where the previous owner was creating theirown electrical wiring.

Not all the wiring is used, and in many cases, live wires areleft without being properly capped or terminated.

Take the time to trace all the electrical wiring to ensurethere are no hidden surprises or weak connections.

Overloaded circuits

Overloaded circuits can occur when additional poweroutlets are created to use existing wiring as their source.

This shortcut method creates problems where multipleappliances are plugged in and drawing power at the sametime.

The demand exceeds the capacity and causes short fuses.

Determine the power drain on each circuit and arrange tocorrect the wiring to ensure that each circuit load isbalanced.

Climatic conditions

Climatic conditions can affect electrical industries on alarge scale.

A slight deviation in the monsoon can affect theproduction in a large way.

So it is essential to obtain clear data before the planningof production of energy.

Intense summer can also be a problem as a largeamount of energy will be consumed during summer.

Overpopulation

With increase in population the average demand haveincreased, which in turn have lead to a need to increase theproduction of electricity.

Population is increasing rapidly which results in an increasein demand.

Overpopulation have caused a hike in electricity productionover the past decade.

Insufficient water

Water is an important element in power generation.

Water is used in hydro power plants for generation, innuclear power plants for cooling purpose etc... sufficientwater availability is always an area of concern for electricalindustries.

Hence it is important that water is conserved with properplanning, so that it can be used effectively when required.

Availability of non-renewable resources

Over use of non-renewable resources in the past haveresulted in the considerable decrease of resources suchas coal, petroleum etc..

Most of the production of electricity is based on non-renewable resources.

Excessive use of these resources in the present cancause decrease in production in the future.

Coal is one of the main non-renewable resources that isbeing exploited, as it is the main raw material inthermal plants.

It is recommended that renewable resources should be used in a more effective, excessive and efficient manner.

Industrialisation

A large amount of energy is being consumed by industries all the year long.

It is essential that enough energy is available to full fill the needs of the industries.

So the electrical industries must be able cop up with this.

Electrical load forecasting methods

• A model or method is a mathematical description of howthe complex elements of a real-life situation or problemmight interplay at some future date.

• In projecting electricity demand, a method uses data onelectricity prices, income, population, the economy, andthe growth rates for each and then varies the mix accordingto varying sets of assumptions.

• Different assumptions produce different outcomes.

• The relationships between electricity demand and themultitude of factors that influence or affect electricitydemand are expressed in mathematical equations calledfunctions.

• A model is a collection of functions. A function, in turn, ismade up of variables for which those factors which changeor can be changed.

• Independent variables are those factors which influencethe demand for electricity, and the dependent variable iselectricity demand itself.

• In other words, the demand for electricity depends onpopulation, income, prices, etc.

• Finally, elasticities describe how much the dependentvariable (electricity demand) changes in sense to smallchanges in the independent variables.

• Elasticities are what the modeler uses to measureconsumer behavior.

• All of the forecasting methods are capable of looking atdifferent scenarios and do so by changing their basicassumptions .

Long Term Load Forecasting Methods

The long term load forecasting methods are basically,

Trend analysis

Econometric modeling

End-use analysis

Combined analysis

Trend Analysis

• The trend extrapolation method uses the information of the past to forecast the load of the future.

• A simple example is shown in figure(2010), in which load is shown for the last 10 years and predicted to be 2906 MW in 2015.

• A curve fitting approach may be employed to findthe load of the target year.

• This approach is simple to understand andinexpensive to implement.

• However, it implicitly assumes that the trends invarious load driving parameters remain unchangedduring the study period.

Econometric modeling• In this approach, initially the relationship between the load

and the driving parameter 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,

• where i denotes the year and a, b, c and d are theparameters to be determined from the historical data.

• Once this relationship is established, the future values ofthe 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 variouscustomer classes (residential, commercial, etc.) and to thesystem as a whole.

• It is relatively simple to apply.

• The drawback is the assumption of holding the relationshipestablished for the past to be applicable for the future.

• In this way, the influence of any new driving parametercannot be taken into account.

• Econometrics uses economics, mathematics, and statisticsto forecast electricity demand.

• Econometrics is a combination of trend analysis and end-use analysis, but it does not make the trend-analyst’sassumption that future electricity demand can be projectedbased on past demand.

• Econometrics uses complex mathematical equations toshow past relationships between electricity demand andthe factors which influence that demand.

• For instance, an equation can show how electricity demandin the past reacted to population growth, price changes,etc.

• For each influencing factor, the equation can showwhether the factor caused an increase or decrease inelectricity demand, as well as the size (in percent) of theincrease or decrease.

• For price changes, the equation can also show how long ittook consumers to respond to the changes.

• The equation is then tested and fine tuned to make surethat it is as reliable a representation as possible of the pastrelationships.

• Once this is done, projected values of demand-influencingfactors (population, income, prices) are put into theequation to make the forecast.

• A similar procedure is followed for all of the equations inthe model.

• Econometric models work best when forecasting atnational, regional, or state levels.

• For smaller geographical areas, meeting the model can be aproblem.

Advantages

• It provides detailed information on future levels ofelectricity demand, why future electricity demandincreases or decreases, and how electricity demand isaffected by various factors.

• It provides separate load forecasts for residential,commercial, and industrial sectors.

• It is flexible and useful for analyzing load growth underdifferent scenarios.

Disadvantages

• For an econometric forecast to be accurate, thechanges in electricity demand caused by changes in thefactors influencing that demand must remain the samein the forecast period as in the past.

• This assumption (which is called constant elasticities)may be hard to justify, especially where very largeelectricity price changes (as opposed to small, gradualchanges) make consumers more sensitive to electricity

prices .

End-use analysis• This type of analysis is mostly confined to residential loads

but may be applied with some modifications to other loadclasses, too.

• As a simple example, if refrigerator is concerned, based onthe number of households and estimating the percent ofhouseholds having a refrigerator, the number ofrefrigerators for a future year may be estimated.

• Based on average energy use of such an appliance, the totalenergy consumption of refrigerators may be estimated.

• It is obvious that the average energy use is dependent onthe intensity of appliance use, its efficiency and thermalefficiency of homes.

• The same procedure may be applied to other type ofappliances and equipment in order to forecast the totalenergy requirement.

• As evident, this approach explicitly predicts the energyconsumption.

• If the load is to be estimated, some indirect approacheshave to be used to convert the predicted energy to load(power demand).

• This approach may lead to accurate results if its extensiveaccurate data requirements can be provided.

• Various driving parameters effects may be taken intoaccount.

• The basic idea of end-use analysis is that the demand forelectricity depends on what it is used for (the end-use).

• Ideally this approach is very accurate. However, it issensitive to the amount and quality of end-use data.

• End-use forecast requires less historical data but moreinformation about customers and their equipment

Advantages• It identifies exactly where electricity goes, how much is

used for each purpose, and the potential for additionalconservation for each end-use.

• End-use analysis provides specific information on howenergy requirements can be reduced over time fromconservation measures such as improved insulationlevels, increased use of storm windows, building codechanges, or improved appliance efficiencies.

• An end-use model also breaks down electricity intoresidential, commercial and industrial demands.

• Such a model can be used to forecast load changescaused by changes within one sector (residential, forexample) and load changes resulting indirectly fromchanges in the other two sectors.

• Commercial sector end-use models currently beingdeveloped have the capability of making energy demandforecasts by end-uses as specific as type of business andtype of building.

• This is a major improvement over projecting only sector-wide energy consumption and using economic anddemographic data for large geographical areas.

Disadvantages

• The disadvantage of end-use analysis is that most end-use models assume a constant relationship betweenelectricity and end-use (electricity per appliance, orelectricity used per dollar of industrial output).

• This might hold true over a few years, but over a 10-or20-year period, energy savings technology or energyprices will undoubtedly change, and the relationships willnot remain constant.

• End-use analysis also requires extensive data, since allrelationships between electric load and all the many end-uses must be calculated as precisely as possible.

• Data on the existing stock of energy-consuming capital(buildings, machinery, etc.) in many cases is very limited.

• Also, if the data needed for end-use analysis is not current,it may not accurately reflect either present or futureconditions, and this can affect the accuracy of the forecast.

• Finally, end-use analysis, without an econometriccomponent that is explained above, does not take pricechanges (elasticity of demand) in electricity or othercompeting fuels into consideration.

Combined Analysis

• The end-use and econometric methods may besimultaneously used to forecast the load.

• It has the advantages and disadvantages of bothapproaches.

• The only way to determine the accuracy of any loadforecast is to wait until the forecast year has ended andthen compare the actual load to the forecast load.

• Even though the whole idea of forecasts is accuracy,nothing was said in the comparison of the three forecastingmethods about which method produces the most accurateforecasts.

• The only thing certain shut any long-range forecast is that itcan never be absolutely precise.

• Forecasting accuracy depends on the quality and quantityof the historical data used, the validity of the forecaster’sbasic assumptions, and the accuracy of the forecasts of thedemand-influencing factors (population, income, price,etc.).

Short term load forecasting methods

• Short-Term Load Forecasting is basically is a load predictingsystem with a leading time of one hour to seven days,which is necessary for adequate scheduling and operationof power and profitable management in electrical utilities,short-term load forecasting has lot of importance.

• High forecasting accuracy and speed are the two mostimportant requirements of short-term load forecasting andit is important to analyze the load characteristics andidentify the main factors affecting the load.

• In electricity markets, the traditional load affecting factorssuch as season, day type and weather, electricity price thathave voluntary and may have a complicated relationshipwith system load..

• Various forecasting methods have been applied to short-term load forecasting to improve accuracy and efficiency.

The short term load forecasting methods are ,

Similar Day Lookup Approach

Regression Based Approach

Time Series Analysis

Artificial Neural Network

Expert System

Fuzzy logic

Support Vector Machines

Similar Day Lookup Approach

• Similar day approach is based on searching historical dataof days of one, two or three years having the similarcharacteristics to the day of forecast.

• The characteristics include similar weather conditions,similar day of the week or date.

• The load of the similar day is considered as the forecast.

• Now, instead of taking a single similar day, forecasting isdone through linear combinations or regression proceduresby taking several similar days.

• The trend coefficients of the previous years are extractedfrom the similar days and forecast of the concern day isdone on their basis.

Regression Based Approach • The term "regression" was used in the nineteenth century

to describe a biological phenomenon.• Linear regression is a technique which examines the

dependent variable to specified independent.• The independent variables are firstly considered because

changes occur in them unfortunately.• In energy forecasting, the dependent variable is usually

demand or price of the electricity because it depends onproduction which on the other hand depends on theindependent variables.

• Independent variables are usually weather related, such astemperature, humidity or wind speed.

• Slope coefficients measure the sensitivity of the dependentvariable that how they changes with the independentvariable.

• The future value of the dependent variable can beestimated.

• Essentially, regression analysis attempts to measure thedegree of correlation between the dependent andindependent variables, thereby establishing the latter’spredicted values.

• Regression is the one of most widely used statisticaltechniques.

• For electric load forecasting, regression methods areusually used to model the relationship of load consumptionand other factors such as weather, day type, and customerclass.

• There are several regression models for the next day peakforecasting.

• Their models contain deterministic influences such asholidays, random variables influences such as averageloads, and exogenous influences such as weather.

Time Series Analysis • Time series forecasting is based on the idea that reliable

predictions can be achieved by modeling patterns in a timeseries plot, and then extrapolating those patterns to the future.

• Using historical data as input, time series analysis fits a modelaccording to seasonality and trend.

• Time series models can be accurate in some situations, but areespecially complex and require large amounts of historical data.

• Additionally, careful efforts must made to ensure an accuratetime line throughout data collection filtering modeling andrecall processes.

• Time series analysis widely used in the martial management forforecasting of customer demand for goods services.

• Time series approaches are not widely used for energy industryforecasting. Because they typically do not take into accountother key factor, such as weather forecasts .

• Time series have been used for longtime in such fields aseconomics, digital signal processing, as well as electric loadforecasting.

• In particular, ARMA (autoregressive moving average),ARIMA (autoregressive integrated moving average), ARMAX(autoregressive moving average with exogenous variables),and ARIMAX (autoregressive integrated moving averagewith exogenous variables) are the most used classical timeseries methods.

• ARMA models are usually used for stationary processeswhile ARIMA is an extension of ARMA for non-stationaryprocesses.

• ARMA and ARIMA use the time and load as the only inputparameters.

• Since load generally depends on the weather and time ofthe day, ARIMAX is the most natural tool for loadforecasting among the classical time series models.

Artificial Neural Networks • Artificial Neural Networks are still at very early stage

electronic models based on the neural structure of thebrain.

• In a neural network, the basic processing element is theneuron.

• These neurons get input from some source, combine them,perform all necessary operations and put the final resultson the output.

• Artificial neural networks are developed since mid-1980and extensively applied. They have very successfulapplications in pattern recognition and many otherproblems.

• Forecasting is based on the pattern observed from the pastevent and estimates the values for the future.

ANN is well suited to forecasting for two reasons:

• First, it has been demonstrated that ANN are able toapproximate numerically any continuous function to bedesired accuracy. In this case the ANN is seen asmultivariate, nonlinear and nonparametric methods.

• Secondly, ANNs are date-driven methods, in the sensethat it is not necessary for the researcher to usetentative models and then estimate their parameters.ANNs are able to automatically map the relationshipbetween input and output; they learn this relationshipand store this learning into their parameters.

To use the ANN in electric load forecast problems,distribution engineers should decide upon a number ofbasic variables, these variables include:

• Input variable to the ANN (load, temperature…etc)

• Number of classes (weekday, weekend, season…etc)

• What to forecast: hourly loads, next day peak load, nextday total load …etc

• Neural network structure (Feed forward, number of hiddenlayer, number of neuron in the hidden layer…etc)

• Training method and stopping criterion

• Activation functions

• Size of the training data

• Size of the test data

Expert Systems • Expert systems are new techniques that have come out as a

result of advances in the field of artificial intelligence (AI) inthe last two decades.

• An expert system is a computer program, which has theability to act as an expert.

• This means this computer program can reason, explain, andhave its knowledge base expanded as new informationbecomes available to it.

• The load forecast model is built using the knowledge aboutthe load forecast domain from an expert in the field.

• The "Knowledge Engineer" extracts this knowledge fromload forecast (domain) expert which is called theacquisition module component of the expert system.

• This knowledge is represented as facts and rules by usingthe first predicate logic to represent the facts and IF-THENproduction rules.

• This algorithm consists of functions that have beendeveloped for the load forecast model based on the logicaland syntactical relationship between the weather andprevailing daily load shapes in the form of rules in a rule-base.

• The rule-base developed consists of the set of relationshipsbetween the changes in the system load and changes innatural and forced condition factors that affect the use ofelectricity.

• The extraction of these rules was done off-line, and wasdependent on the operator experience and observations bythe authors in most cases.

• Statistical packages were used to support or reject some ofthe possible relationships that have been observed

• The rule-base consisted of all rules taking the IF-THEN formand mathematical expressions.

• This rule-base is used daily to generate the forecasts.

• Some of the rules do not change over time, some changevery slowly while others change continuously and henceare to be updated from time to time .

Fuzzy Logic • Fuzzy logic based on the usual Boolean logic which is used

for digital circuit design.

• In Boolean logic, the input may be the truth value in theform of “0” and “1”.

• In case of fuzzy logic, the input is related to the comparisonbased on qualities.

• For example, we can say that a transformer load may be“low” and “high”.

• Fuzzy logic allows us to deduce outputs from inputslogically.

• In this sense, the fuzzy facilitate for mapping betweeninputs and outputs like curve fitting .

• The advantage of fuzzy logic is that there is no need ofmathematical models for mapping between inputs andoutputs and also there is no need of precise or even noisefree inputs.

• Based on the general rules, properly designed fuzzy logicsystems are very strong for the electrical load forecasting.

• There are many situations where we require the preciseoutputs.

• After the whole processing is done using the fuzzy logic, the“defuzzification” is done to get the precise outputs.

• The use of these intelligent methods like fuzzy logic andexpert systems provide advantage on other conventionalmethods.

• The numerical aspects and uncertainties are suitable forthe fuzzy methodologies.

Support Vector Machines • Support Vector Machines (SVM) are the most powerful and

very recent techniques for the solution of classification andregression problems.

• This approach was come to known from the work ofVapnik’s, his statistical learning theory.

• Other from the neural network and other intelligentsystems, which try to define the complex functions of theinputs, support vector machines use the nonlinear mappingof the data in to high dimensional features by using thekernel functions mostly.

• In support vector machines, we use simple linear functionsto create linear decision boundaries in the new space

• In the case of neural network, the problem is in thechoosing of architecture and in the case of support vectormachine, problems occurs in choosing a suitable kernel.

• Mohandes applied a method of support vector machinesfor short-term electrical load forecasting.

• He compares its method performance with theautoregressive method.

• The results indicate that SVMs compare favorably againstthe autoregressive method.

• Chen also proposed a SVM model to predict daily loaddemand of a month.

Spatial Load Forecasting

• Planning for the future expansion of a power systeminvolves determining both the capacities and the locationsof future components, namely, generation facilities,transmission/ sub-transmission/distribution lines andcables and various substations.

• This requires forecasting the future loads with geographicdetails (locations and magnitudes).

• In power system context, this topic is addressed as spatialload forecasting.

• Spatial load forecasting is accomplished by dividing utilitysystem into a number of small areas and forecasting theload of each.

• In some cases, the small areas used may be irregular inshape or size, corresponding to the service areas assignedto particular delivery system components such assubstations or feeders.

• A simple choice is to use square cells that cover the regionto be studied. Once the load of each cell is predicted, theelectric load of the system (or a larger geographical area)can be predicted.

• An important aspect of electric load is that cells (smallareas) do not simultaneously demand their peak powers.

• The coincidence factor defined as the ratio of peak systemload to the sum of small area peak loads is, normally in therange of 0.3–0.7

• If we focus on a small area, is it really possible to predictGDP and population rate for a small area.

Multivariate forecasting methods

• Multivariate forecasting methods rely on models in the statistical sense of the word, though there have been some attempts at generalizing extrapolation methods to the multivariate case.

1. Is multivariate better than univariate

In forecasting, and even in economics, multivariate modelsare not necessarily better than univariate ones.

While multivariate models are convenient in modelinginteresting interdependencies and achieve a better (notworse) fit within a given sample, it is often found thatunivariate methods outperform multivariate methods outof sample.

Multivariate models have more parameters thanunivariate ones.

Model selection is therefore more complex andlengthier and more susceptible to errors, which thenaffect prediction.

It is difficult to generalize nonlinear procedures to themultivariate case. Generally, multivariate models musthave a simpler structure than univariate ones, toovercome the additional complexity that is imposed bybeing multivariate.

2.Static and dynamic forecasts

Dynamic forecasting is that there have to becalculated forecasts for the periods after the firstperiod in the sample simply by using the previouslyforecasted values of the lagged left-hand variable.

Static forecasting uses actual rather than forecastedvalue for the lagged variable and which can be doneonly if there are actual data available.

3. Cross correlations An important exploratory tool for modeling multivariate

time series is the cross correlation function (CCF).

The CCF generalizes the ACF(Autocorrelation function) tothe multivariate case.

Thus, its main purpose is to find linear dynamicrelationships in time series data that have been generatedfrom stationary processes.

4. Single-equation models

Two types of so-called single-equation models can beconsidered for multivariate forecasting: regressionmodels and transfer-function models.

Both types are ‘open-loop’ models and model a dynamicrelationship of an ‘endogenous’ variable that depends onone or several ‘explanatory’ variables.

The methods are helpful in forecasting only if futurevalues of the explanatory variables are known, ifforecasting of explanatory variables is particularly easy,or at least if there is no dynamic feedback from theendogenous to the explanatory variables.

5. Vector auto regressions and VARMA models

• These models usually treat all n variables in a vectorvariable y as ‘endogenous’.

6. Co integration

• When there is cointegration, cointegrated models shouldbe used for forecasting.

• When there is no cointegration, series with an integratedappearance should be differenced.

7. Multivariate state-space models

State-space models, such as Harvey’s ‘structural’ time-series models, can be generalized to the multivariate case.Chatfield draws particular attention to the SUTSE(seemingly unrelated time-series equations) model withcommon trends.