The FINAL 416 Paper Team Ironwood

8/4/2019 The FINAL 416 Paper Team Ironwood

1/38

Hydro Generator Simulator

Final Project Report

Team Ironwood

Angel Barahona-Sanchez

Craig Bjorklund

Matt Colby

David Fugate

Yonas Woldemichael

EE 416

Washington State University

Team Sponsor: Avista Utilities

Mentor: Kristina Newhouse

Instructor: Scott Campbell


2/38


3/38


4/38


5/38

money and time by giving them the option to test their equipment in house. A possiblesecondary market includes schools and training environments, where operators andtechnicians can learn from a hands-on learning tool.

One major revision to this project has taken place between the two semesters of

this senior design course. At the end of the first semester (EE415), it was this groupsintent to have both a software simulator (Phase I) and a physical interface panel (PhaseII) completed by the end of the second semester (EE416). However, realizations fromboth the group and Avista determined that completing both phases of the project wouldbe too cumbersome. Therefore, the hydro generator software simulator (Phase I) is thefinal object to be presented to Avista at the end of EE416. The interface panel, Phase II ofthe project, will not be completed by this design team but a starting point for itsintegration with the simulator will be provided.

The five students responsible for the design and implementation of the hydrogenerator simulator remain unchanged from last semester and are known as team

Ironwood. This team consists of five electrical engineering students, Angel Barahona-Sanchez, Craig Bjorklund, Matt Colby, David Fugate, and Yonas Woldemichael, all intheir senior year at Washington State University.

Each member will contribute evenly to the efforts of the group, but to maximizeefficiency, the project has been broken into different parts and assigned to groupmembers. There are four main portions that define the following project plan: Marketing,Quality, Engineering, and Manufacturing. The following list shows the division of roleswithin these portions as shown in table 1:

Table 1. Divisional RolesMarketing MattQuality AngelEngineering

Intro DavidGenerator AngelGovernor/Turbine MattProtection YonasExcitation System DavidHMI & Simulator Matt & CraigPhase II Craig & Matt

Manufacturing Craig & Yonas

5


6/38

II. Marketing:

Market Analysis:

The primary customer of the hydro simulator is Avista Utilities. Avista desires the

ability to test control device settings in a safe environment without risking damage toexpensive equipment such as a generator. In the past, engineers from Avista had to travelto Colorado to test a governor on a Woodward generator simulator. Although thissimulator provided an adequate testing environment, resources were spent on traveling toColorado from Spokane, WA. The governor was also damaged in shipping resulting inunnecessary downtime. An in house generator simulator would provide Avista with acost effective solution.

The market for the hydro generator simulator is primarily suited towards powerutilities that use hydro electric dams to generate power. Since few hydro generatorsimulators exist in the U.S, the primary market for the generator simulator is fairly large.

The Federal Energy Regulatory Commission (FERC) regulates more than 1,500 hydroelectric dams in the U.S. alone [1]. Power generation utilities must provide a cheap,reliable source of power as efficiently as possible with minimum downtime. Failure to doso could result in power outages, loss of money, and wasted resources. In order to meetstandards, adequate control schemes will need to be developed and tested. A hydrogenerator simulator provides an effective solution for testing device settings.

A possible secondary market could be trade schools and training for powergeneration control operators. It is essential for power generation operators to have handson experience before working in an actual control room. The hydro generator simulatorcould provide the means to meet this end. In a class room or lab environment students

would have the opportunity to apply theoretical knowledge to a simulation of an actualhydropower plant. This would allow students to make changes to different systemparameters without worrying about damaging expensive equipment in the event ofunexpected performance.

Marketing Features:

The hydro generator simulator is developed in National Instruments Labview 8.2.Labview provides the software interface that can be used both to manipulate inputs to thesimulator and monitor the system. Using this software the simulator is modeled to reflectthe necessary equipment for the control and operation of a hydro power generation

system. A hydro turbine, generator, excitation system, governor, and protection elementare programmed to reflect a real world system. These models respond according to thespecifications given in the WECC and IEEE standards.

The ingenuity of the simulator is that with Labviews simulation module the usercan interact with the simulation during its real time operation. Control inputs can beadjusted while the software is running and the responses can be readily seen. The user no

6


7/38


8/38

Documentation necessary to get started implementing the interface panel is alsoprovided. Being that this is a custom designed panel for each customers specifications,the layout of the product will vary slightly from product to product. To aid in anyconfusion that may exist because of this, documentation for the panel will be included onthe same disk as the simulator software. This documentation includes mappings of what

control device signals attach to the DAQ modules and which attach to the PLC.

III. Engineering:

Introduction:

The development of the hydro generator simulator was broken into two phases.Phase I was to develop a virtual simulator while Phase II was to gather information onhow to interface real devices with the simulator. Phase I was further broken down intofive sections which makes up the simulator: the generator, governor/turbine, excitationsystem, protection device, and the human machine interface (HMI). In each section theprocess used to develop the model is explained. First, an explanation of what the modelrepresents in the real world is covered. Next is a description of the different parts of eachmodel and how they interact with each other. Following this is an explanation of how themodel was programmed and validated using Labview. Problems that were encounteredwhen developing the particular Labview model are explained.

3.1 Generator Model:

Background:

Synchronous generators form the principle source of electrical energy in todayspower systems and are the core component of the generator simulator. In producing sucha model for the generator, an understanding of its characteristics and accurately modelingthe dynamics are of fundamental importance. Described below is a brief description ofthe synchronous generator and its function, followed by the research done in order tochose an appropriate model.

The synchronous generator consists of a round or salient rotor. Its type dependson the rated rotational speed of the turbine. For example, at higher speeds, typicallyaround 1800 rpms and greater, a round rotor is used; and at lower speeds, a salient polerotor is used. Typically round rotor generators are found on steam powered units, and

salient pole rotors on hydro powered units.

Another important part of a generator is the stator, which consists of armaturewindings that are distributed 120 apart in space. This is done in order to produceuniform rotation of the magnetic field when excitation current is applied to the fieldwinding of the rotor. Because of this, the three phase voltages in the armature windingsare displaced by 120 in time.

8


9/38

The production of three phase voltage is the result of the total net magnetomotiveforce produced by the field and armature windings and there relation according toFaradays Law which is shown below.

(1) (2)

In the above equations, see Equation (1) and Equation (2), ei is the inducedvoltage, the instantaneous value of flux linkage at time t,L the inductance,R theresistance of the conducting wire, and i the per unit current entering the circuit in Figure1. Because of the large amounts of circuits, like Figure 1, that are involved in thesynchronous machine, and the fact that the mutual and self inductance of the statorcircuits vary with rotor position synchronous generator equations are complicated [2].This complication poses a problem when describing the electrical performance ofsynchronous machines, or when modeling a synchronous generator. Historically this hasbeen always a challenge.

Figure 1. Single-excited magnetic circuit [2]

In the past, to simplify the modeling and analysis of transients for a synchronous

generator, two axes were defined. The direct axis, or d-axis, was centered magneticallyon the north pole of the rotor, and the quadrature axis, or q-axis, was 90 electrical degreesahead of the d-axis, as shown in Figure 2. The d and q axis were further developed bymathematically transforming the 3 phase stator quantities into corresponding two axisquantities. These transforms developed by R.H. Park, are the widely known Parkstransformations. The effects of such transformations are to move all the machine time-varying inductance coefficients from the machine flux linkage equations. The widespreaduse of the direct axis and quadrature axis equations has been developed from theseconcepts. These can also be visualized in terms ofdand q axis equivalent circuits. Theorder can be defined simply as the number of rotor circuits in either the d or q axis;depending upon the number of inductance/resistance series combinations representing the

field and direct axis equivalent rotor circuits, or the number representing quadrature axisequivalent circuits [3].

9


10/38

Figure 2. Stator and rotor circuits of a synchronous machine [2]

Choosing a Model:

The basic approach taken in generating or finding a model was to first locatehistorical work done in this area. The IEEE Xplorer library was the primary source thatprovided information relating to model parameters and test procedures of synchronousgenerators. Specifically the IEEE standards 1110-1991 and 115-1995 provided a basicunderstanding of this subject. After reading through the IEEE standards, many conceptsrelated to properly modeling the synchronous generator were still unclear

These modeling uncertainties created difficulties in finding the correct model for

the generator. Even though much work has been done historically on modeling dynamicsin a synchronous generator for stability studies, assistance was needed to put thisinformation into some context. Individuals like Larry Long [16], Professor Tomsovic[18], and Professor Mani [19] provided assistance in understanding the concepts neededto find the correct model for the synchronous generator and its relating equations.

When choosing a model for the synchronous generator the main considerationwas keeping to the specifications of Avista. The model desired was a salient pole

10


11/38

generator. Being that most salient pole generators are constructed with laminate rotors,the rotors usually include copper damper bars located in the pole faces. The purpose ofthe damper windings is to reduce the mechanical oscillations of the rotor aboutsynchronous speed [4]. These damper bars tend to form a squirrel-cage amortisseurcircuit that is effective both in the direct and quadrature axes because they are connected

at their pole faces with continuous end-rings. Since this amortisseur is the only physicalcircuit present in the q-axis, a first-order model can describe it adequately. Hence, Model2.1 is recommended for most salient pole generators [3].

The model chosen was the gensal model from the WECC approved modellibrary because it represents a salient pole synchronous generator. This model is shown inFigure 3 and is represented as model 2.1 in Figure 4. The main reason it differs from theround rotor model is that throughout each revolution of the rotor, the self inductances ofthe stator, and the mutual inductances between them, are not constant. These values varyas a function of the rotor angular displacement , which is shown in Figure 2 [4].Theangle , is the angle between the axis of phase a and the d-axis [2]. Also, the salient-pole

model doesnt saturate significantly in the quadrature axis as it does in the round rotormodel, and thus no quadrature axis saturation is present for Figure 3. As a result, theequations for the flux linkages of the salient-pole machine are more difficult to use thantheir round-rotor counterparts [4].Fortunately, the equations that pertain to the a, b, and cphases can be transformed by using Parks transformations.

++

+

++

-

+

+ +

-

id

-

-Efd

+-

iq

-

-

d-axis

q-axis

Figure 3. Linearized synchronous generator block diagram of a salient pole, gensal

11


12/38

Rfd

R1d

L1dLad

Ll Lf1d

L1q

R1q

Laq

Ll

Figure 4. Model 2.1

Model Description:

The model shown above in Figure 3 has been the standard salient-pole modelused in recent years for small-disturbances. The direct axis sub transient open circuit timeconstant, T`d0, and the direct axis sub transient open circuit time constant, T`d0, are usedto account for time responses of faults that might occur. For example, if a terminal faultwas simulated from an open circuit, the stator currents would be inversely proportional toXd0 (for the sub transient period) and inversely proportional toXd, (for the transientperiod) during the fault [3]. These time constants are normally between 0.01 and 9.0seconds, depending on the type of generator[2].

When the terminal voltage and three-phase currents of the machine change, sodoes the field voltage of the rotor windings. This typically causes saturation in the d-axisof the rotor. The generator open-circuit saturation curve in the model shows thissaturation relationship and is used to determine the saturation factors in this axis. Aninternal voltage behind some specified reactance is used to locate the operating pointon the open-circuit saturation curve to calculate a saturation factor K (or a saturationcorrection E) [3]. Thus, the internal excitationXaduifd, (orLaduifd) is then the sum ofseveral components [3]. This form of excitation determination has had widespread usesince the early 1960s [3]. Equation (3) shows this relationship to Figure 3 in per unit.

EXXIEEiX dddqIfdad ++== )(``

(3)

12


13/38

Programming in Labview:

Using the simulation module in Labview made programming of the synchronousgenerator model rather straightforward. A snapshot of the synchronous generator is inFigure 5 below illustrating the manner this was done. The complications that did arise

occurred during the verification stage of the model.

Figure 5. Screenshot of salient-pole model in Labview.

Due to the fact that the generator is part of an integrated system which includesvarious controls, the synchronous generator, by itself, could only be verified to a limitedextent. These tests included responses to changes in field voltage inputs and changes tothe respective input axis currents. The results were used to verify that a particular changein input resulted in the correct output from the model. After these basic tests werecompleted, the generator model was connected to the excitation model of the simulator.

Once connected with the excitation model, a general test was conducted to insurethat it worked properly. This was accomplished by changing the reference voltage in theexciter model and observing if the terminal voltage of the generator responsded correctly.After this, the synchronous generator model was incorporated into the simulator with aload and circuit breaker so validation tests could done. Tests were evaluated bycomparing the synchronous generator outputs from the Noxon report [5] to the simulatorsynchronous generator outputs. The only discrepancy between the simulator and the

13


14/38

Noxon report was that a round rotor synchronous generator was modeled for the Noxontesting rather than a salient-pole generator.

3.2 Governor/Turbine Model:

Background:

Speed governors are used to regulate the output torque and rotational speed of theprime mover, in this case the hydro turbine. By controlling the energy input to theturbine, the governor can effectively regulate the MW load on the generator in acontinual effort to match the generation to the load. By controlling the rotational speed ofthe turbine the governor can control the electrical frequency of the generator. Toaccomplish these two tasks the governor adjusts the turbine wicket gates, which affectsthe amount of water that flows into the guide vanes of the turbine.

It was desired by Avista for the governor model to represent an electro-hydraulic

governor. This type of governor uses solid state electronics to implement feedbackcontrol. It senses the speed of the turbine using a frequency transducer and converts thissignal into a DC voltage using a frequency to voltage converter. This signal is comparedwith reference DC voltage that represents the desired operating frequency. The errorbetween these signals is fed into the PID controller. The output of the governor is ananalog signal that operates the control mechanism necessary to adjust the wicket gateposition. For example, if the measured electrical frequency is lower than the desiredfrequency the governor gives a raise signal to open the wicket gates. This increases theshaft speed and therefore the electrical frequency.

It was also necessary for the model to include speed droop. Speed droop opens thecontrol valve a specified amount for a given disturbance. By definition droop is thepercent difference between the no-load and full load speeds of the unit. [6] It is necessaryfor generators to operate with stability in parallel with other generators in interconnectedsystems. In operation, droop causes the generation unit to decrease speed for anincreasing power output so that it doesnt continually work to maintain a constant speedunder different loading conditions. This allows for generators to share a load increase inproportion to the different ratings of the generators in the system.

Most electronic governors implement droop that receives feedback directly from apower (watt) transducer from the generator potential transformers (PTs) and currenttransformers (CTs), instead of the control gate position. This is known as speedregulation [6]. Speed regulation is preferred over droop with gate position feedbackbecause it is not affected by the nonlinear relationship between the gate position and thewater flow. Electro-hydraulic governors also include a speed reference signal. Using thisinput signal an operator or an automatic control system can increase the desired rotationalspeed of the unit. The speed reference signal can be used to set the individual loading of aunit.

14


15/38

Governor/turbine models also include representation of the hydraulic systemwhich is used to position the wicket gates. In the real world this is usually a pressurizedoil system. A pilot valve is used to direct pressurized oil to the prime mover actuators ascontrolled by the PID controller. The pressurized oil is relayed to servo motors thattransmit this hydraulic pressure to a rotating gate ring which is attached to the wicket

gates of the turbine. The rate at which the gate opens and closes as well as the timeconstants of the pilot value and servo motors need to modeled in order for an accuratesimulator.

In addition to the governor and hydraulic system it was also necessary for a primemover to be modeled. The prime mover converts kinetic energy from rushing water intomechanical energy necessary to rotate the shaft of the generator. A Francis turbine ismodeled as the prime mover. A Francis turbine is a reaction turbine that is used at damswith larger head levels, usually 50 to 2400 feet [7]. How the Francis turbine performs isinfluenced by several characteristics of the turbine as well as the water column behind theturbine. Several non-ideal characteristics of the water column include water inertia and

water compressibility as well as the elasticity of the penstock walls. Head level, flow rate,and penstock length are also factors that should be accounted for. These factors areusually modeled into the water starting time. The water starting time is defined as thetime required for water flow to in the penstock to accelerate from zero to no load watervelocity given some initial head level. [2]

The turbines relationship between the ideal gate opening and real gate openingwas also necessary to be modeled. This is modeled as an overall turbine gain. This can befound by inverting the difference between the full load gate position and the no load gateposition. Finally the non-linear relationship between the gate position and power at theturbine must be modeled. This relationship is measured during the real time operation ofthe turbine and modeled using a lookup table.

Choosing a Model:

The process of researching and finding an appropriate governor/turbine modelwas fairly difficult. Initially, during the research process, the main source of informationwas the IEEE Xplorer. The IEEE Xplorer is a library of IEEE documents that cover awide range of technical subjects which includes power system control modeling. Many ofthe governor models found were designed to represent the older mechanical governorswhich included transient and permanent droop instead of PID control and speedregulation.

The initial model chosen for the simulator was the PID governor model gpwsccas recommended by Larry Long [16]. This is an accepted model of the WesternElectricity Coordinating Council (WECC) and can be used in most steady state andtransient analysis studies. The only drawback is that the turbine depicted in this modelwas an ideal lossless turbine represented by a single transfer function. For simulationsinvolving large variations in power output and frequency this ideal turbine model is notappropriate [2]. It was preferred to have a more accurate turbine model that accounts for

15


16/38

non-ideal properties such as an inelastic water column as well as flow rate and head level.After contacting Kristina Newhouse [17], the model Avista provided was the hyg3 modelas shown in Figure 6. This model is used to represent the governor/turbine for Unit 1 atNoxon Rapids generation station.

Figure 6. Hyg3 Model

Model Description:

As seen from Figure 6 the desired governor characteristics (droop, PID control,etc.) are incorporated into this model. Droop is shown as relec while the PID controllergains are ki, k1 and k2. Transducer delay effects are also modeled using first ordertransfer functions. The inputs to the model are electrical power, reference power (or

speed setpoint) and the speed deviation. The resulting outputs are the mechanical powerand the gate valve position.

It should be noted that the model shown in Figure 6 has been modified from theoriginal Hyg3 model. The model shown in Figure 6 neglects the option of using gatedroop feedback from the control value. Instead this model only allows for electrical droop(speed regulation) because it is more frequently used. Although deadband is shown in

16

Governor

Hydraulic system

Francis Turbine


17/38

Figure 6 it is usually not modeled because it is difficult to get the necessary data to modelit [2].

A hydraulic system and detailed turbine model representing a Francis turbine arealso shown in hyg3. For the hydraulic system, the first transfer function represents thepilot valve and servo motors. This transfer function is limited by the maximum rate at

which the gate can open and close. The gate position is modeled by the integrator blockand is limited by the maximum and minimum gate positions. The output of the hydraulicsystem in the model is the gate valve position.

The value of the gate value position is fed into the non-linear lookup table thatconverts the gate position to the power at the gate valve. This value is then fed into theturbine model which calculates running values of the head level and flow rate. The watertime constant, as discussed earlier, is modeled as Tw. The block labeled At is used toaccount for the difference between the ideal gate opening and real gate opening.


Programming the hyg3 model into Labview using the simulation toolkit was fairlystraightforward. Since the simulation toolkit supports linear and non-linear functions suchas transfer functions and saturation limiters, the block diagram shown in Figure 6 wasprogrammed in a few hours. A screen capture of the governor/turbine model is shown inFigure 7 below.

Figure 7. Screenshot of the governor/turbine model in Labview

17


18/38


19/38


20/38

calculate a compounded terminal voltage. With load compensation, the excitation systemcan be used to regulate voltage at a set point out in the system. Although currently notrelevant for the hydro simulator, some excitation systems include an automatic voltageregulator, AVR, which uses inputs from other local synchronous machines to control themachine's terminal voltage since they share a common load.

The fourth subsystem is the power system stabilizer, or PSS. The PSS provides anadditional signal to the regulator to enhance damping of power system oscillations.Commonly used input signals are terminal voltage, accelerating power, rotor speed, andfrequency deviation. Although the power system stabilizer is an important part is systemstability it is not implemented at Noxon unit 1 therefore is not necessary to be modeled.

The fifth subsystem is the protective circuits and limiters. Various control andprotective features work to ensure that the limits of the both the exciter and the generatorare not exceeded. If they are exceeded, some of these functions can also take emergencyaction and signal the breaker to take the generator offline. Some of the commonly used

functions are over/under excitation, field-current limiter, terminal voltage limiter, andvolts/hertz limiter.

Excitation systems need to be able to control the generators operation not onlyduring steady state conditions, but during transient, and post disturbance conditions.Failure to due so could result in power outages and equipment damage. The ceilingvoltage and the response rate are the main factors that determine how well an excitationsystem can respond to sudden changes during disturbances.

The ceiling voltage is the maximum field voltage the exciter can operate at [7]. Itusually ranges from 1.5 to 6 times the field voltage during full load conditions. Theresponse rates deals with how fast the excitation system can respond to changes involtage. It is defined as the time required for the exciter to go from open circuit voltage tothe ceiling voltage when then generator is at full load field voltage [7]. These conditionsare necessary to be incorporated into the model.

Choosing a Model:

The excitation model to be used in the hydro simulator is EXAC8B, an approvedWECC model, which is also known as the AC8B by IEEE. This model is shown in Figure10, with the colored blocks representing the different sections of the model. TheEXAC8B was chosen after Larry Long [16] suggested using this model due to the factthat it reflects the digital PID controls that are similar to the Basler DECS300. TheEXAC8B models an AC excitation system which uses an AC alternator and eitherstationary or rotating rectifiers to produce the DC field. Although this model wasdeveloped for brushless AC exciters, it can be used to model a DC exciter simply bysettings the gains of Kc and Kd to zero.

20


21/38

Figure 10. AC8b/EXAC8B Model

The AC type models are not valid for frequency deviations of +- 5% from the

rated frequency, 60Hz and oscillation frequencies up to about 3 Hz [8]. This is becausethe model does not account for regulator modulation as a function of the systemfrequency. Hence this model should not be used to study sub-synchronous resonance orother shaft torsional interaction problems. The synchronous machine's field current mustbe supplied back to the model in order for it to represent loading effect accurately.

Model Description:

EXAC8B receives the following inputs from the synchronous generator:compound terminal voltage Vcomp (or Vc), generator field current Ifd, and Vs fromthe power systems stabilizer, if in use. Section A in Figure 10 is the PID voltage regulator.

This takes the sum of Vsig, Vc, and Vref, and amplifies the signal in block B. Thetime constant, Tdr represents lag from the PID controls and Ta is the lag from thevoltage amplifier. The constant Ka represents the voltage regulators set gain. Theoutput of the voltage regulator is now a regulated voltage, Vr. This signal is used tocontrol the excitation system, which is modeled in blocks C and D.

Section C uses Vr in a feedback loop and also receives the generator fieldcurrent, Ifd. The time constant in the integration block is the lag associated with theexciter. The output from this block is the exciter voltage, Ve. This voltage is multipliedwith Fex to represent Efd, the exciter field voltage that is fed to the generator. Theexciter voltage is used to calculate a voltage, Vx, which is proportional to exciter

saturation between the field current and field voltage as the load increases.

Section D models rectifier regulation. Rectifier regulation is a non-linear effectthat decreases the rectifier output voltage as the load current increases. The expression for'Fex' is determined by the value of 'In.' There are three specific modes of operation, asshown in Table 2.

Table 2. Rectifier regulation effect

21


22/38

Mode 1 f(In)=1.0+0.577In In 0.433

Mode 2 f(In)= 2In-0.75 0.433 < In < 0.75

Mode 3 f(In)=1.732(1.0-In) 0.75 In 1.0

It should be noted that although DC exciters ignore effects of rectifier regulation and fieldcurrent feedback, these features were still modeled incase a user desired to simulate anAC excitation system.


By using Labviews simulation toolkit, programming the excitation system tookonly a couple of hours. Once EXAC8B was programmed into Labview, a few basic testswere performed to validate that the model worked as expected. Model parameters fromthe Noxon report [5], specifically unit 1, were programmed into the model and set as themodels default parameters. Figure 11 shows a screen capture of the programmed

excitation system model.

Figure 11. Screenshot of the excitation system in Labview

To perform the test, the output Efd was fed back to the input, Vc. Normally thecompound terminal voltage, Vt from the generator would be fed into Vc, but since

Efd and Vt are proportional it can be done. The gains Kc and Kd were set to zero forthis to model a DC excitation system. The additional voltage signal, Vsig was set tozero because this input is from a PSS and this component is not needed for modeling unit1 of the Noxon plant. For the first test, the reference voltage, Vref, was initially set to0.1 per unit (pu) and speed to 1 pu. When Vref was increased to 0.8 pu, Efd alsoincreased. When Vref was decreased back to 0.1 pu, Efd decreased, see Figure 12.This test demonstrated that when the reference voltage is increased or decreased, the PIDcontroller drives the field voltage to this same reference value.

22


23/38

Figure 12. Resulting field voltage after changing the input, Vref

For the next test Vref was held at 0.5 pu and the speed signal was changed.When speed was decreased from 1 pu to 0.8 pu, Efd initially decreased. The voltageregulator amplified this error and the PID controller brought the level of the field voltageback up to the reference value. Likewise, when the speed signal was increased, Efdincreased and then automatically decreased to the set level, as shown in Figure 13. Theseresults follow what should be expected. If the speed of the generator decreases, poweroutput and terminal voltage will decrease as well. To stay at the desired voltage theexcitation system compensates for the decrease in speed by increasing Efd. The sameresponse is seen if the speed increases, but instead Efd is decreased to maintain thereference voltage.

Figure 13. Resulting field voltage after change in speed

A few problems were encountered while working on the excitation model. Thefirst problem with the model was that the original diagram provided by Larry Long wasincorrect. Highlighted in Figure 14, block A has two inputs but no outputs. This problemwas thought to be solved by looking at other documentation of the EXAC8B [9]; which

23


24/38

shows that the output of the block A connects directly to block B and that the only inputto block A was from Ladifd, not from block C.

Figure 14 . Exac8b model with labeled blocks

Eventually, while looking at the two different sources, it was discovered thatblock A actually received inputs from block C and Ladifd, [2],[8],. It was also verifiedthat block A output was the input to block B. The rest of the model was verified to becorrect. The other problem with the EXAC8B model was some confusion in what blockB represented. It was later discovered that the output from block A, representing therectifier current, which is used to determine which equation should be used to calculateFex, see Table 2.

3.4 Protection Model:

Background:

A protection model was also necessary be included in the simulator. This modelwas developed using several features from the Schweitzer 300G protection relay. The300G provides protection for the generator by controlling the current, voltage, andfrequency outputs from the generator. Some of the key features of the 300G relay includecurrent differential protection, out-of-step protection, over excitation detection,directional power protection, volts/hertz Protection.

Even though the relay provides numerous protection features, only a handful offeatures were chosen to be incorporated into the simulation design due to the design ofthe generator model. For example, in the simulator, the generator output is simply a perunit value for the magnitude of the terminal voltage. Since many of the protectionelements use a three phase input for their calculations, only the over and under voltage,over and under current and frequency protection elements were chosen to be modeled.

24


25/38

Over and Under Voltage:Phase over voltage protection operates by measuring the output voltage at the

generator terminals and compares this with a reference maximum value. In an actual unitthe voltage is simply measured with potential transformers (PTs). If the measuredsecondary voltage is higher than the maximum reference voltage, the under voltage

protection element is activated. This element will open the circuit after a certain amountof time proportional to the magnitude of the measured voltage, if it is greater than thereference voltage. Phase under voltage protection uses this same concept.

Over and Undercurrent:Phase overcurrent protection operates using the maximum of the measured phase

current magnitudes. Phase undercurrent protection operates using the maximum of themeasured phase current magnitudes. There are many types of overcurrent elements whichthe 300G relay provides for protection of the generator. Some of the typical elements ofprotection are definite time overcurrent, neutral time overcurrent, residual timeovercurrent, and voltage controlled definite time overcurrent protection.

Frequency Protection:Over frequency conditions usually occur during dramatic load variations. In order

to provide protection during these situations frequency protection is provided. The relayprovides six bands of over/under frequency protection. The pickup settings for the underand over frequency elements are 59.5 Hz and 60.5 Hz respectively. When the frequencydeviates below 59.5 Hz it will fall in one of the six protection bands. The greater thefrequency deviation the fast the relay will operate. Figure 15 shown below provides amore detailed description of how under frequency protection works in the 300G relay.

Figure 15. Under frequency operation [10]

The band between 60 and 59.5 Hz is the area of unrestricted timeoperatingfrequency, while the dotted areas below 59.5 Hz are areas of restricted time operating.Table 3 shows a more detailed description of the different operating conditions.

25


26/38

Table 3. Under frequency operation description [11]Frequency Band (Hz) Time delay Comments60-59.5 - No action, generator can

operate59.5 and below 1.5 s Continuous under frequency

alarm59.558.858.8- 58.058-57.557.5-5757 - 56.556.5 and blow

50 min9 min1.7min14 sec2.4 sec1.0 sec

Alarm under frequencylimit exceeded. Thesebands may trip or alarmdepending on individualutilities practices.


As with the other models, Labviews simulation toolkit was used to implement

the protection model. One challenge during programming the model was making thecorrect timer from the simulation toolkit for the frequency and overcurrent elements.Initially, the frequency and current protection models failed in situations when thesimulation speed was increased because the timer used in the while loop and simulationloop were mismatched. The simulation loop timer ran faster than the while loop timer.Therefore, a different approach was used where the timing from one loop was linked tothe other.

Over and Under Voltage Labview Model:The function of the under and over voltage model is to read the terminal voltage

from the generator and feed this value to a comparator with respect to the voltagereference. The comparator then checks if the given voltage is below or above thereference voltage. A Boolean value is returned to determine if the relay is to trip thecircuit breaker. Figure 16 shows the Labview screenshot for the under and over voltagemodel.

Figure 16.UnderOver

voltagemodel

26


27/38

Overcurrent Labview Model:

The overcurrent element was programmed using the U.S Inverse curve U2. Thisequation can be used to calculate the operating time of the relay see Equation (5).

+=1

95.5180.0*

2

M

TDtp (5)

The time dial (TD) settings are used to control the slope of the inverse curve. Themore inverse the slope, the faster the relay will operate for a given current. The operatingtime (tp) is the time which the relay will operate for a given current input. Multiples ofpickup current (M) are given by the ratio of the input current to pickup setting. In thismodel Ipickup is set to 5A secondary current. Current input (Iinput) is the outputcurrent from the generator. A screen capture of this model is shown in Figure 17.

Figure 17. Screenshot of the Overcurrent model in Labview

Under Frequency Labview Model:The under frequency model monitors the generators operating frequency to

make sure it is within the acceptable bounds of 60 Hz. If the generator's frequency isbelow a certain limit frequency, then this model determines if the generator is

disconnected from the grid. A screen capture of the under frequency model is shown inFigure 18.

27


28/38


29/38

actual Woodward 505H governor in the simulation, the software model for the governorwould be deactivated and the physical devices inputs and outputs would be connected tothe simulator via a PLC and DAQ. The simulation would then run as it normally would,only the governor control signals are given from the actual governor. Figure 19 is shownto give a general schematic of how the physical system would be interconnected.

Figure 19. Phase II layout for connecting a 505H

For purposes of this project, the group has determined the following I/Os to bethose covered to meet the requirement of physically connecting the governor andexcitation system into the simulation. The information given below is based on the I/Ostypical for a Woodward 505H digital governor and a Basler DECS300 voltage regulator.The highlighted items are those covered by the PLC with the remaining to be covered bythe DAQ.

Table 4. I/O Layout between the PLC and DAQ

Governor: Voltage Regulator:(6) Analog Inputs (4-20 mA) (1) Generator Voltage Sensing (~120 V)(6) Analog Outputs (4-20 mA) (2) Generator Current Sensing (~1 A)(1- 2) Speed inputs (1-30 Vrms) (13) Switching contact inputs (24 Vdc)(16) Discrete Inputs (18-26 V dc) (1) Remote set point control Input (-10 to 10 V)(1-2) Actuator Outputs (4-20 mA) (1) Analog output (-10 to 10 volts or 4-20 mA)(8) Discrete Outputs (form C) (8) Contact outputs

29


30/38

In order to handle the various I/Os required of the PLC, various control cards areneeded. To ensure the correct parts were chosen, the customer service department for thePLC was contacted and the above list of requirements was given to them. Avistarequested the use of a Modicon Quantum PLC as it is one that they have used in the past.The company responsible for the Quantum PLC is Schneider Electric, based in Palatine,

IL. The customer service representative (Joe Cyr) was briefed on the requirements of thePLC for this project and the following parts list was established to do the job. The pricingfor the below PLC parts, shown in Table 5, were quoted on 10/22/2007 from Graybar, thesupplier Avista will most likely be using should they choose to order these parts.

Table 5. PLC QuoteDescription Part Number Price

10 Slot Backplane 140XBP01000 $481.24AC Power Supply 115/230V 140CPS11420 $994.47Quantum 434 Controller (High End) 140CPU43412A $8,615.57Quantum 512K CPU (Mid Range) 140CPU11303 $5,178.24

Relay Out 16 x 1 NO 140DRA84000 $740.20Analog Out 4ch Current 140ACO02000 $1,479.61Discrete DC Input Module 140DDI35300 $740.20

Total with HighEnd CPU:

$13,051.29

Total with Mid-Range CPU:

$9,613.96

While these pieces of equipment cover the requirements of the PLC, there are acouple of signals that still remain uncovered for the voltage regulator I/Os. Thegenerator voltage and current sensing signals are two inputs into the voltage regulatorthat require a voltage and current amount that exceeds the rated handling of any PLC andDAQ card available. A possible solution to this may lie within using an external sourcesuch as a Doble power source to generate these high values required. Being that Phase IIis beyond the scope of this groups project, the details in solving this issue will be left asis and remain to be solved by others.

Data Acquisition:

The Labview simulator is designed to also communicate with the physical controldevices using Data Acquisition methods developed by National Instruments, the samecompany that develops Labview. These types of equipment are designed to seamlesslyinterface directly with any virtual instrument developed in Labview with little additionalprogramming. DAQ devices can accept a wide range of analog and digital inputs fromthe control I/Os that will be hardwired to the device itself.

After deciding which control device I/Os will interface with the PLC and whichwill interface with the DAQ, it was desired to find an effective, low cost solution. Aspreviously discussed the PLC will be handling more of the digital I/Os while the DAQwill handle much of the analog I/Os such as the actuator signals and the speed signals.

30


31/38

National Instruments offers several methods that would work. Figure 20 shows several ofthe data busses available.

Figure 20. Bandwidth vs. Latency of various busses [12]

The first option researched was PXI/PCI interface method. This option proved tobe one of the best methods. It could cover most of the required I/Os and also interfacedwith the computer using the PCI slot. Also, the I/O card purchased for this system wouldall run on the same clock so there would be no problems with synchronizing the differentI/O modules. Even though this option had the most benefits, they did not outweigh thecost of this option. After configuring a complete system to handle the necessary analogI/Os, the PXI method would have run roughly $6000 to $8000.

After further research, the method eventually decided upon was the compact DAQsystem. This system provides a simple means to interface with the computer using USB2.0. It is much cheaper than the PXI interface method. As it can be seen from Figure 20,USB 2.0 provides good bandwidth with a fairly low latency. Another benefit in usingUSB is that it is designed to take advantage of the plug and play feature of a Windowscomputer with no additional hardware necessary, similar to a USB thumb drive. Thecomputer can automatically detect the new device when it is attached. The only drawbackto this system is that the USB 2.0 bus shares its bandwidth across all the I/O modules.This means that when all the modules are connected to control devices there may be

some time delay in transferring the data. However according to a National Instrumentsengineer, Avinash Harjani[15], this is not a significant problem for this application. Atypical compact DAQ system is shown below in Figure 21.

31
mailto:[email protected]:[email protected]:[email protected]


32/38

Figure 21. A compact DAQ system attached to a computer [13]

As it can be seen from Figure 21, the compact DAQ uses a chassis where severalI/O modules are connected. The chassis provides power for the signal generation of theI/O modules. The individual control and measurement modules are designed to interfacewith control devices using screw terminals. The process of selecting the appropriate I/Omodules was fairly straightforward. The analog and digital signals from the Woodward505H and Basler DECS300 not covered by the PLC were matched to the appropriatemodules using the compact DAQ builder on the National Instruments website. Afterdeveloping a compact DAQ system a National Instruments applications engineer verifiedthat the I/O modules selected would be appropriate for this application.

Some of the analog signals created more of a problem. One example is the speedsignal that is fed into the Woodward 505H. This speed signal is of the magnitude 1-30volts RMS. The I/O modules for the compact DAQ system are only in the range of 0 to10 volts and 0 to 20 mA. Not only did a DAQ device need to be able to generate thissignal, it had to do so at the appropriate frequency because this signal represents the

signal read from the speed transducer. By working with the applications engineer fromNational Instruments the PCI-6624 was chosen as shown below in Figure 22 .

Figure 22. The NI-PCI-6624 DAQ card [14]

This card plugs directly into the PCI slot of a standard computer. Although thiscard runs at about $1,300, it is able to provide eight channels of frequency controlledanalog voltage. When connected and programmed properly, this DAQ card should beable to mimic the speed signal from a real hydro plant. After verifying this card, a fullquote was developed by customer service at National Instruments. This quote is shownbelow in Figure 23 with the total price for the DAQ system at $4,671.

32


33/38

Figure 23. National Instruments DAQ quote

IV. Quality:

Quality in Software:

When looking at the main work that was done for this project, the majority of thetime was spent within the software models. The generator, exciter/voltage regulator,turbine/governor, and the protection device each have their own Labview softwaremodel. However, the design for each model was based on the general operating principlesof its physical counterpart and the respective standard that it followed. Therefore, thequality of each developed software model was tested to meet similar specifications.

33


34/38

To meet these specifications, each model underwent an end item verificationprocess throughout the design and upon completion of the finished model. This was doneto reduce any final troubleshooting time as well as to verify that the developed model metthe requirements for its respective device after major programming changes were made.Not only was the model checked against requirements of the device, but also verified to

meet the functionality of parameters desired by the customer. With each model passingits respective tests, the final deliverable (hydro generator simulator) was tested to verifythat each component worked compatibly with one-another.

To ensure the quality of the software models, various test procedures specific toeach model were also developed. These verification procedures were implemented foreach respective model, and for a couple of protection schemes. These tests determinedhow accurately the developed software models simulated the operation of the physicaldevices they were representing. An example of such a test was the comparison to theNoxon Report simulation [5], which was done at one of Avistas current facilities. Thiswas emulated on the simulator to see if the modeled equipment responded in a similar

way that the Noxon report did. To expand the testing even more, the simulationparameters were tested to determine the range of the respective parameters the modelscould operate in.

With the equipment models being tested in this fashion, certain parameterboundaries were enabled. The software was made to simulate a hydro generator, soproper operating boundaries were checked. These boundaries were checked before andduring the simulation. For example, if an actual turbine had a speed limitation of 100RPMs, then a user input of 200 RPMs would be responded to with an error promptnotifying the user, telling him to input another value. With these boundaries set andproper testing complete, the quality of each software model was assured.

Reliability:

Finally, to insure the reliability of the software, all specifications of eachrespective model were tested and challenged. This enabled the design team to checkwhen the software was vulnerable to errors, thus rendering it useless. To further prolongthe life of the software, a modular development approach was taken. This approachallowed the design of the software models to be easily substituted in and out of theoverall system. This permitted for future device changes and upgrades to be accountedfor. For example, if a company would change their voltage regulator to a new model, thecorresponding block could be changed by simply adding in an updated module in theplace of the current one.

34


35/38

V. Manufacturing:

Producing a Deliverable:

As a product that is individually tailored for the client, the hydro generator

simulator is an item that is produced in relatively small numbers. The product isdeveloped for power utilities that have specific requirements and regulations that theirpersonalized simulator must adhere to. With the client base of the product being a fairlynarrow one, the client can rest assured that their product will fit their needs. Thispersonalization and attention to detail therefore means that a large manufacturing facilityfor mass production will not be necessary.

The overall manufacturing process for the hydro generator simulator consists oftwo parts, the software and the interface panel. The software consists of the hydrogenerator simulator along with all of the control device models. This single program runsusing the Labview application on the customers computer to form the overall simulation.

The interface panel is the physical device that allows the customer to integrate actualcontrol elements into the computer simulation. The final design and assembly of theinterface will be done by the customer, but information provided by team Ironwoodoffers a starting point and some general information needed to create the interface.

The completed product includes all of the software, interface panel (Phase II)documentation and a manual covering the details for each model and what conditions thesimulator can model. All of this data is included on a compact disk. This results in a verysimple manufacturing process since all that is needed to be produced is a CD.

Product Validation:

With the final software and documentation loaded onto a compact disk, finalverification procedures were implemented to ensure the disk is ready to be delivered tothe customer. The software was first installed onto a generic computer that met minimumspecifications. Once the software was installed, several basic simulations were run. Thisensures that no essential modules of the software are missing or erroneous. The differenttrials place specific stress on individual sub-functions within the software.

Along with the software install validation, the included documentation waschecked to confirm they have been successfully installed on the computer. Since theseprocedures simply verify the proper installation of the software and corresponding

materials, minimal time is taken. After the manufacturing tests have been run, the diskcan then be labeled and is ready for delivery to the customer.

Estimate of Other Production Costs:

As was discussed in the Phase II portion of the Engineering section of this paper,two quotes were obtained covering the costs of the PLC and DAQ equipment. The final

35


36/38

pricing for these items are given in Table 6, neglecting the necessary man-hours neededto assemble the interface panel.

Table 6. Estimated cost of the interface panelHigh Cost Low Cost

Modicon Quantum PLC $13,051.29 $9,613.96NI DAQ $4,671.60 $4,671.60Total $17,722.89 $14,285.56

36


37/38

VI. References:

[1] Wikipedia, Hydropower, www.wikipedia.org, 2007. [Online]. Available:http://en.wikipedia.org/wiki/Hydropower. [Accessed: April. 9, 2007].

[2] [Kundur] P. Kundur, Power System Stability Control, 1st ed., San Francisco:McGraw-Hill Inc., 1994.

[3] IEEE Power Engineering Society, Power System and Electrical MachineryCommittees, " IEEE Std 1110-1991," IEEE Guide for Synchronous Generator ModelingPractices in Stability Analyses, 1991.

[4] J.J. Granger, W.D. Stevenson, Jr., Power System Analyses, 1st ed. , San Francisco:McGraw-Hill, Inc., 1994.

[5] Hannett, Louis N, Report to: Avista Corp. for Noxon Rapids 1-4, 28 April 2005

[6] WECC Control Work Group, WECC Tutorial on Speed Governors, WECC, 1998.

[7] Avista Generation Staff,Introduction to Generation, Avista Utilities, 2001.

[8] IEEE, IEEE Recommended Practice for Excitation System Models for PowerSystem Stability Studies, IEEE Std 421.5-2005 (Revision of IEEE Std 421.5-1992), 21April 2006.

[9] Pacific Gas and Electric Company, System Impact Study, www.energy.ca.gov,2005. [Online]. Available:

http://www.energy.ca.gov/sitingcases/humboldt/documents/applicant/afc/Volume_02/Appendix%205/HBRP_Appendix_5B_System_Impact_Study.pdf, pp. 146 [Accessed: Oct.13, 2007]

[10] Schweitzer Engineering Laboratories Technical Staff, SEL-300G MultifunctionGenerator Relay Instruction Manual, Schweitzer Engineering Laboratories, 2006, pp. 2-99.

[11] IEEE Power Engineering Society, Power System and Electrical MachineryCommittees, ANSI/IEEE C37.106-1987,"An American National Standard IEEE Guidefor Abnormal Frequency for Power Generating Plants, 1987, pp. 19

[12] National Instruments, Digital What Makes a Bus High Performance, www.ni.com,2007. [Online]. Available: http://zone.ni.com/devzone/cda/tut/p/id/4819. [Accessed: Sep.17, 2007].

[13] National Instruments, NI Compact DAQ, www.ni.com, 2007. [Online]. Available:http://www.ni.com/dataacquisition/compactdaq. [Accessed: Nov. 15, 2007].

37


38/38

[14] National Instruments, NI PCI-6624, www.ni.com, 2007. [Online]. Available:http://sine.ni.com/nips/cds/view/p/lang/en/nid/12501. [Accessed: Nov. 15, 2007].

[15] Avinash Harjani (private communication). 2007.

[16] Larry Long (private communication), 2007.

[17] Kristina Newhouse (private communication), 2007.

[18] Kevin Tomsovic (private communication). 2007.

[19] Mani Venkatasubramanian (private communication). 2007.

Documents

The FINAL 416 Paper Team Ironwood