 Welcome to class 16 on topics in power electronics and distributed generation. We have been talking about active anti-alending in the class and last time we looked at the relationship between power and voltage and we modeled the feeder as a parallel RLC resonant circuit and looked at the relationship between power and voltage and then how you could add power voltage characteristics to cause the voltage to have not a stable operating point but can actually go out of the nominal range. Then we started looking at the reactive power versus frequency relationship and we are able to get a relationship this should be the relationship between reactive power. So, this is Q the relationship between reactive power and frequency related to your quality factor power and frequency and at the nominal 50 hertz your power of the injected by the d g is 0 and we saw that if your frequency deviates by delta f say reduces by delta f then your reactive power drawn by the load would increase by delta q. If your frequency increases by delta f your the power drawn by the load would reduce by amount of again some delta q load but with a negative polarity. Then we looked at what would happen if you had variations in L and C around the nominal values and if your essentially your q of your load increases that would correspond to the situation where your L reduces. So, drawing more reactive power or C also reduces. So, this would correspond to then a situation where your resonant frequency will now settle down to a higher value or the on the other side if your q load changes in a polarity where your q reduces that would correspond to L prime going to L plus delta L and some C prime being C plus delta C and the corresponding resonant frequency that it would settle down to is at a lower value. And if you look at what would be the operating point with the nominal reactive power being 0 output from the d g that would be essentially a flat line and you could look at then what would be the final operating frequency depending on what the final resonant frequency of that particular system is. Then you could look at what is the next situation here we looked at where the there is variation in L and C around the nominal value you could then look at what would happen if your there is a change in reactive power from the d g. And here what we are assuming is that the d g power variation is modeled as a equivalent L C load variation and essentially if once your upstream switch opens in the on the feeder then essentially under that condition after switch S 1 opens you have your q load is equal to q d g because there is no delta q coming in at that particular point from the model of the system that you are trying to analyze. So, q load is what is being consumed and q d g is what is being sourced at the d g point. And if your d g operating q shifts by a value of delta q d g. So, this would be your original value and if it shifts up then essentially you could think of it as a now operating at a equivalent as a equivalent load with what we saw was it would be L minus delta L or C minus delta C. So, you would settle down at a higher operating frequency. So, similarly if you have a drop in reactive power then essentially your operating frequency would actually come down to a lower value. So, it is possible to shift the operating frequency of your island based on adjusting the amount of reactive power that you are injecting from your d g. So, if you look at it from a way on what you would do to destabilize the feeder you could link your reactive power to change in frequency that is measured at the d g. So, essentially what you do is change your output of your d g with some gain times q q quality factor power operating frequency with delta f. So, you could think of essentially your block diagram as under nominal conditions q d g command would be 0. You could measure the frequency and pass it through a high pass filtered or a wash out network to get the change in frequency. And you can link that with k q f and provide this and depending on the type of d g that you have whether it is a synchronous machine or a inverter you could link it to the exciter. So, in a synchronous machine you can control your output bar by adjusting the excitation level in a power converter you would can control the amount of reactive power output by controlling your quadrature access current which your in phase current would correspond to real power your quadrature access current would correspond to the reactive power. So, by linking in a manner in this particular manner you could think of shifting the operating frequency. So, the idea is if you have a disturbance which would cause say your frequency to lower. So, as an example if you have a case where your frequency dips then essentially what you do is you lower the war output from your d g and the lowered war output from the d g would cause a further dip in your frequency. And this is further measured in measured at your sensing point leading to further reduction in frequency and essentially your system operating frequency goes out of bound. So, one thing that you can see in this particular analysis that we did for your feeder along with the distributed d g source in a case where you have unintentional islanding is that and compare it with what you expect from a traditional power systems analysis book. In a traditional power systems analysis book the link is between power and frequency and between voltage and reactive power. So, whereas that is not the case in what we have analyzed so far the main reason is that the model that we take for the feeder is different from the model that is considered typically in power systems analysis. If you look at the large transmission systems, the large generators, the major portion of the loads in a power system is actually machine loads more than 60 percent of the power consumed actually goes into large motors. So, if you look at that the model in that particular case is a machine with large inertia. So, there the link is between the energy and speed of the machine where the model is the of the power going to accelerate or decelerate the inertia. And then the other thing is that you look at the lines the equipment they are predominantly inductive. So, you have see transition lines with high x by r ratios. So, you can consider that to be primarily inductive. So, in a situation such as that the link would be different from what you get in the model of a feeder where we have taken the model of the feeder to be a parallel RLC resonant load. But overall the objective and the case where we are trying to detect an unintentional island is to make your operating point unstable. So, if you have say your voltage amplitude versus time and at some particular instant t naught say your upstream switch open. So, the idea is to have a destabilizing term such that your voltage goes out of bound. So, you have a v high or a v low which would be the thresholds at which your d g would detect and disconnect. And depending on whatever it is you look at the ability of your anti islanding algorithm to be able to detect in a short duration that your voltage has gone out of bound or if you have frequency you are looking at your frequency versus time you have say a nominal frequency and at t naught you your upstream switch disconnected. And then you have a f high or a f low and depending on when depending on when you are able to detect that the frequency went out of bound you can trigger the relay at your d g source that something has your your nominal condition has gone beyond your range your operating condition has gone beyond acceptable range. So, the d g has to disconnect and this duration that you typically have is less than 2 seconds because you want to be able to detect that unintentional island has occurred before upstream recloser recloses. So, you want this detection to happen fairly quickly. So, the disconnection can happen anywhere upstream of the on the feeder and the d g can be anywhere located anywhere along the feeder after the disconnection the unit which is actually making the detection of whether it is islanded situation or not is detecting its voltage at locally it is not able to detect at some other point. So, it is taking that measurement on a local basis and based on the local measurement it is making a decision whether to stay connected or to disconnect. So, in literature there are a variety of algorithms that people have looked at methods that people have looked at for both passive and active anti islanding. We looked at under and over voltage under over frequency we also looked at reverse power flow based detection people have looked at other methods such as power factor change rate of change of power rate of change of frequency whether there is a sudden phase jump on in your system whether there is a large imbalance that is happening on in the in terms of the voltage whether there is a increase in T h d that is being measured. So, some of these methods would be considered passive because here you are just observing whereas in the active methods you are trying to introduce changes such as actively trying to change your operating frequency or change your power level voltage level etcetera. So, it goes by a variety of names, but you now have a feel for what is a underlying principle that people are trying to adopt essentially you are trying to take stable operating point and make it unstable. So, that your actual voltage is measured and if you see that your power levels or voltage amplitudes or frequency goes out of bound that would correspond to a situation of unintentional island. Also you there are always advantages and disadvantages for many of these schemes and to find improved methods for detecting an unintentional island is a area of active research. So, one thing that you would have seen is that from the point of view of detecting and responding to a fault on your feeder and if you are for a variety of reasons you would need to disconnect from your feeder and you need to do that fairly quickly. And couple of primary reasons would be to disconnect from the grid for protection coordination, protection and coordination also to prevent formation of unintentional island. And there is another reason why you would like to disconnect rapidly from the grid and this is where you want to operate as a intentional island. So, this is primarily for power quality reasons. So, if your grid power quality is poor for some reason and your switching speed is slow it means that your load is now exposed to that poor power quality for a longer duration. So, if you can rapidly disconnect from the grid it means that the duration of poor power quality seen by the load can be reduced by a faster switch. If you look at then a traditional electromechanical circuit breaker your operating duration is 2 to 5 cycles and that would be considered instantaneous. So, the fastest disconnection might be of the order of multiple cycles whereas, if you look at semiconductor based switch you could switch in a much faster manner in less than a cycle whereas, if you look at a semiconductor based switch. So, if you look at then a situation where you have the grid potential distributed generation source and then switch which can connect or disconnect then you have two possibilities. So, you could have the grid so you might from a power quality perspective you might be normally under normal condition you might be connected to the grid and if you have a situation where your grid power quality is poor or the grid goes away you have a blackout essentially you want to switch over and connect to a DG. Then in terms of operation of the DG for power quality perspective there are two possible ways in which you could operate your DG you could have say a cold standby which means that your DG is deenergized and it needs to start up before. So, normally the grid voltage is there if you need to change over first you need to start up your source before you can actually transfer over to the source and this would be typically the way you would run a diesel generator set. And if you look at the start up time of gen set it can take 20 seconds for a really fast start up gen set to minutes for something which might need a little bit more warm up time to actually start up and stabilize to a normal operating condition. So, one possibility is in response to the grid power quality being poor you then start up a gen set then the duration of outage seen by the load would be longer of the order of then 20 seconds to a minute depending on what is your start up time another way of running at would be a hot standby. And if you look at essentially a situation such as that it means that your voltage over here is available all the time. So, that would correspond to a situation where for example, you could have the machine spinning at under no load or it could be a UPS which because the power loss in a UPS can be much smaller compared to the power loss in a gen set your UPS is always having a output voltage or many UPS can actually start up in a much shorter duration depending on how quickly you can start up your inverter. So, that would correspond to a hot standby basis which means that essentially you have voltage available and in this particular case the speed with which you are able to transfer over is then entirely limited by the speed of the switch. So, here you are not waiting for a DG to start up if your switch is faster you can actually transfer over in a much faster manner. . So, if you look at power quality applications so use of electromechanical and semiconductor based transfer switches are commercially available solutions. So, there are commercial transfer switches that are both solid state and electromechanical. So, then if you look at a semiconductor based switch what you and compare it with say electromechanical switch then you can look at what would be your advantage and what would be the penalty that you would have. So, you can look at and if you look at the speed definitely the electromechanical switch is much faster if you look at the number of cycles your semiconductor based switch can operate much larger number of times compared to a mechanical circuit breaker because after given number of operations may be 10,000 20,000 cycles you might have to look at your springs and do servicing of your breaker. Whereas your semiconductor based switch can operate for millions of cycles and without much degradation. Then if you look at the other aspects you look at power loss the power loss in a semiconductor based switch is much higher because of conduction loss on state drops etcetera. So, your electromechanical switch is actually more attractive from power loss perspective if you look at it in terms of electrical ruggedness your electromechanical switch can take much more abuse in terms of surge current surge voltages etcetera compared to semiconductor based device. So, there is definitely advantage of electromechanical compared to the semiconductor based switch in terms of electrical ruggedness cost definitely the semiconductor based switch is actually much going to be much more expensive. So, there are tradeoffs. So, you can see cost is a important factor for many commercial applications. So, if you are willing to pay the cost it means that your load has some definite advantage by the improvement in power quality that the electromechanical the semiconductor based switch can provide just if there was no net overall cost advantage then there would not be an incentive to go for a semiconductor based switch. So, then in terms of the semiconductors you could then look at what are the possible semiconductors if you look at the earliest devices that were available in solid state those would be the power diodes the S C R etcetera the thyristors. So, the S C R or the silicon controlled rectifier would be a early device that was that is that has been available since the 1960s. So, comparatively compared to other devices it is a mature technology compared to I G B T is transistors etcetera it has much higher surge current rating. So, if you look at the peak to nominal current rating of S C R it would be higher much higher than that of an I G B T also if you look at in terms of the on state drop which has a implication for power loss the on state drop of an S C R is actually lower than that of typical transistor. So, of similar rating so S C R based semiconductor switch has been some other thing a particular technology that people have looked at for now quite a while. So, if you look at S C R so it is it can be thought of as a switch where to turn on you apply a pulse between your gate and the cathode when you have positive voltage across the anode and cathode then the device would turn on a S C R cannot turn off by itself. So, you need a external circuit to turn it off in a AC system you have voltages going through your currents going through 0 crossings. So, you have a points where your currents go through 0 and you could make use of that to actually stop conduction within say a cycle or so. And for turning off you would need to keep your current in the S C R to be below the holding current of the device for a duration greater than its turn off duration and with no gate pulses applied. So, that would be what is required to turn off the device and after it is off to turn it back on you have to apply a gate cathode voltage pulse when you have positive anode to cathode voltage. So, what we mentioned was a S C R has higher surge current capability. So, if you look at the higher power S C R they are commonly available in what is called as a press pack packages. So, essentially the press pack packages have couple of advantages that are beneficial one is it is possible to have very high current devices. So, the press pack packages essentially what is like a plate a cylindrical plate and essentially now one side would be your anode the other side would be your cathode and you would have leads for gate and cathode that is coming in from the side. So, to hold a press pack devices device in place you apply pressure on the top and the bottom and now that would mean that you have conduction of power loss from two directions compared to one direction for a traditional say I G B T module that is mounted on a heat sink. So, your thermal resistances can that can be advantages in terms of thermal resistances also because you are applying pressure from the top and bottom of such a device you the fail a fail device would typically fail short because any failure would cause the device the local junction to get damaged and the damage junction would essentially then short between your anode and cathode. Whereas, in a module type of device the modules are connected with wire bonds and if for some reason the module ruptures then there is a possibility that the rupture can cause the failed module to fail open in which case you can have a failure that can either be short circuit or open circuit in case of a heat sink mounted module. Whereas, in a press pack package when you have failure it is typically a short circuit which means that now if you have a device that fails you can have a series connection of multiple devices and you can have redundancy because the failed device is short the next device can always be used to actually control. So, for many high power applications where redundancy is a important concern you could then consider a press pack package which will give you benefits. One disadvantage of the SCR compared to something like the modern day IGBT is that SCRs need snubbers for protection. So, you have dv dt snubbers to prevent spurious triggering spurious turn on also you have you need over voltage protection if there is a surge voltage to prevent that surge voltage from triggering inadvertently triggering your SCR you need voltage protection also you need dv dt protection because once you turn on if suddenly a current large current flows through a device in a rapid manner you can have current crowding overheating of information of hot spots and that can lead to damage of the device. So, you with SCRs you would need to use snubbers for its operation. An example of a SCR base transfer switch is shown over here. So, what is shown is essentially a single line diagram of a transfer switch and you have two sources the first source might be considered a prime source. So, under normal conditions your power flow would be from your prime source when the prime source for some reason if the power quality becomes poor you switch over to a secondary or an alternate source and you have in typical transfer switches you might also have some additional switches for normal operation and for bypass operation. For example, under normal operation your switch S 1 n S 2 n and S 3 n would be on your bypass switches S 1 b S 2 b would be off and whether your power is flowing from source 1 or source 2 depends on which set of triistors you are triggering. You might have a bypass operation where S 1 n S 2 n S 3 n is off and S 1 b or S 2 b is on depending on which source you want to bypass to. So, you might make use of this bypass switches to say for example, if you want to repair your main SCR's and still provide power to your critical load or you want to do some modifications on your alternate source you might typically operated in a bypass mode and then ensure that your what was servicing repairs etcetera occurs during a bypass mode and under normal mode you are able to switch between your prime source or your alternate source. And if you look at what the this arrangement is trying to do under normal conditions your power flow would be from V 1. So, your S 1 f and S 1 r would be triggered under normal condition and if for some reason the power quality at V 1 becomes poor if the voltage drops or if it gets gets to distorted or if the some parameter goes out of the acceptable range for the critical load then you want to shift over to V 2. But there are few methods for shifting over to V 2. So, methods of transfer one is called open transition in open transition essentially it means that to transfer from one source to the other first you open both the switches and then there wait for a small delay to ensure that the switches are perfectly open then you turn on the switch to which you want to transfer to. So, here you are having a break before make operation and here the delays can be longer and the power or voltage would be seen by the load for a longer duration. The other possibility is a closed transition in a closed transition essentially you turn on both the switches first and then you turn off the switch you want the source which you want to disconnect from. So, here this is make before break and here if the voltage of the sources are not well matched then there is a possibility for surge currents between your two sources. So, it would the closed transition will not work under all situations or you might end up tripping a breaker or some other switch. Also you could have say something like a forced transition where if you have a primary source and a alternate source and your alternate source can be controlled and it can be controlled in such a manner that you could make use of the voltage and current capability of your alternate source to actively force commutate your outgoing SCR then you could have a transition in a shorter time frame. So, if you look at typical situation you might use a combination of say of these methods. For example, when you are going from prime to alternate you might use open transition because the prime to alternate transition is because you might have a outage in the main power. So, if you do a closed transition you will have a surge of power between your two sources, but when you go from alternate to prime you might have a closed transition which means that now your grid has returned back to normal and you want to turn off from your alternate source and return back to the grid because now both are in the nominal range you can actually connect the two sources together without having a big surge current and then with minimal disturbance to the load you can actually achieve the transfer. Also you could think about a situation where you have a possibility of making use of one source to commutate the second source. So, say if the load is at unity power factor. So, if you assume it is a resistive load and v 1 and v 2 are in phase then you could think about say if you are operating in the positive half cycle of the voltage and if under that particular condition if the prime source voltage went away because of an outage then essentially your if you look at the switch that would be conducting under that condition it would be s 1 f is a positive half cycle s 1 f is conducting current because the resist load is unity power factor. So, this voltage has say gone down to 0 and now this voltage is also in phase which means that this is also seeing a positive voltage. So, if you can turn on s 2 f then you could reverse the apply a negative bias to your thyristor s 1 f and be able to turn off the thyristor by making use of your alternate source. So, in this particular manner you could actually get faster transitions and then you could look at what would be the worst case scenario what would be the maximum delay possible and typically what people have seen is that with transfer switch you can have a quarter cycle delay. So, people talk about a transfer switch of quarter cycle speed and a challenge in many of this transfer switch is to differentiate when to transfer and when not to transfer because you can have many false troops. For example, you might have some small phase jump or some smaller amplitude changes or some harmonics which might be acceptable. So, how to make a decision when to transfer and when not to transfer that is actually a challenge in transfer in such a configuration. So, when you need to definitely make a decision to go across and when not to because if you are thinking about a large load you are thinking about transferring a large amount of power from one source to another source. So, there are consequences of shifting large amounts of power and you do not want to minimize the number of false change overs. The third requirement of such a transfer switch is that you have a switch where you always have the possibility that you might have short circuit in your load and your switches have to be able to handle the high current levels when your load itself has a failure which means that the switch should be capable of handling not just the nominal load current, but also the short circuit currents that the load might encounter. Even if it is for a limited duration of may be one cycle or two cycle, the semiconductor should not get damaged in that short duration. So, designing these transfer switches are quite challenging people if you look at the design of a transfer switch you have two sources if you then look at people also talk about static breakers here you do not now do not have an alternate source to help with your commutation which means that it is even more challenging your duration and delays might be longer than quarter cycle it can be 1 to 2 cycles. So, configuration of a semiconductor based circuit breaker is shown over here where you have the semiconductor based breaker with switches for normal operation and switch for bypass operation depending on how we want to operate your system. And again one main concern in the system is how to handle the peak requirements, the peak in terms of how to handle the fault current requirements and also the continuous power loss that would happen in the semiconductor device when it is operating under normal conditions. People have also looked at hybrid approaches of trying to combine both electro mechanical and semiconductor based switches to minimize your power loss, but have the rapid disconnection capability. So, there are a variety of approaches such as this that are available, but again you have the issue of handling the power require the peak power which would then reflect back in terms of the junction power loss and also the increase cost of now having just a simple breaker now being replaced by a more complicated system consisting of semiconductors plus the other switches for normal operation different modes of operation. So, in a example such as this you could have a semiconductor based breaker and you have the main grid coming in. So, in case there is a poor power quality or the need for detection of an unintentional oil and on the feeder then you can rapidly disconnect from the grid using a semiconductor based switch. Your DG would then continue to provide power to your critical loads, you might have some secondary loads which are not that important which you might be able to say it can actually have a blackout or a outage, but you might have some important loads which are critical which you want to ensure that it stays up even when the power quality goes down. So, this also shows an example configuration where you are using a DG not just for providing power to the main grid exporting power, but also then simultaneously to provide power quality for your critical loads in your facility. So, for references related to these issues you have IEEE standard 1547. So, here this is a recommended standard it is just a recommendation it is not enforceable, but then you have similar issues that are shown in things like UL 1741 which is a underwriter lab standard. So, if you need a UL seal on your equipment which is required for maybe financing etcetera you would need to meet that particular standard. So, many times recommended standards become indirect financial incentives to actually follow and achieve that particular target. This is a one reference which is available from the general electric website on relaying. So, many of these issues reflect back on relaying also you have the journal articles on intentional islanding. Thank you.