Use of a Frequency Response Metric to Assess the Planning and Operating Requirements

so we’ll go ahead and get started so welcome to the first I for energy seminar of spring to 2013 and welcome also to those from the other campuses UC Davis UC Merced and UC Santa Cruz we’re very happy today to kick off this this semester with talk from Joe Edo so I will introduce him Joseph H Edo is a staff scientist at the UC sorry the Lawrence Berkeley National Lab where he is the leader of the electricity markets and policy group and the strategic advisor for the energy storage and demand resources department Joe also leads the program office for the consortium of for electricity reliability technology solutions which is a National Laboratory university industry Rd consortium that conducts research and analysis on electricity reliability and transmission Joe has authored over 150 publications on electricity reliability transmission congestion power quality demand response distributed generation energy technology market transformation utility integrated resource planning and demand-side management and building energy efficiency technologies in 1989 he received both the Crosby field award for best technical paper and the Willis each carrier award for the best presentation by an author under the age of 32 from a Schrade Joe received an a/b and philosophy of science and an MS and energy resources from the University of California Berkeley so he’s one of our own he’s a registered mechanical professional mechanical engineer in the state of California and with that please join me in welcoming Joe ed oh you’re not supposed to say what date I receive that award because now people will know how old well thank you very much Teresa thank you all thank you for having me be protect a part of the I for energy of a series and this is a long overdue presentation because actually I was supposed to get this last year when gaming approached me to give this talk and my schedule is too nutty to do so and I’m very pleased that he’d come back to see or gret what he asked for so I’m going to talk about a project that I completed for the Federal Energy Regulatory Commission a couple of years ago and for those if you don’t already know the Federal Energy Regulatory Commission now has Authority for enforcement of reliability rules across the nation and at the time they are that they asked us to do this study they are very concerned about the rapid growth in renewable energy particularly the bulk power system particularly the form of wind and energy so when I talk about variable renewable energy generation I’ll talk mainly about wind power and they were concerned about you know this rapid growth and in wind power and whether it might cause a reliability threaten so they asked us to do this study specifically to focus on this question of frequency response and so what I’ll do is talk about what came out of that study in kind of a selected fashion their study was a large report in and of itself I won’t talk about all of that report I will share with you at the very end the large cast of supporting technical reports and the website we can get at them if you have interest in probing some of these topics further so first I’ll try to motivate the talk for those of you not familiar with some of these concepts I’ll give kind of a power system frequency control 101 in a couple of slides I’ll then talk about how increased variable renewable generation affects reliability from the standpoint of affecting the frequency response of interconnections I’ll then talk about some of the dynamic simulations with studies we conducted of each of the three interconnections in the United States to try to prove out some of these concepts and prove out the usefulness of these metrics and I’ll spend most of my time talking about the recommendations this work was completed two years ago and I spent most of my time in there in very intervening two years trying to implement or move forward with some of the recommendations so a lot of them kind of date on stuff that’s happened since the worklist finished and then of course I’ll talk about some of the resources that you could access to get more information so this I think is known by many people well renewable energy has gotten a very big push at the federal legislative level through the production tax credits and talk last year of a national RPS and certainly indirectly through talked about carbon legislation

there’s a lot of state mandates directing the acquisition of more renewable power wind is among the lower cost of those sources and it’s currently predominant there are dramatic growth in the installation of wind capacity across the country but integrating that win and putting it into the power system is not straightforward because the wind blows where people don’t live that’s why they don’t live there and so if you want to use the power from the one you have to get it from where they live to one where the wind blows to where people live and then you have to deal with the fact that the wind doesn’t always blow and so from a transmission infrastructure perspective you know you got to build a lot of lines to connect the wind this is a huge institutional challenge I spent a lot of my time supporting the Department of Energy trying to improve the state of into regional transmission planning it’s not a technical issue so much as it is an institutional issue of being able to build powerlines to deliver the wind what I will focus on here is on operating the power system with lots of wind and dealing with the intermittency that that wind poses which is fundamentally different from all the other generation we have on the power system which is controllable which is dispatched to a particular set point or power operating point so an important part of the research that’s done these days is focusing on the forecasting of wind what is the variability that you have to manage but really the when it comes down to is you have an ensemble of units some that are more controllable than others and you have to sort of orchestrate the symphony to get them all to singing the same tune in order to meet load yeah every second of the day and that’s really the focus of what I’m going to talk about is how you manage that synth that ensemble of players to deal with this one particular player that sort of doesn’t watched its own drummer but follows its own rules in terms of what it’s going to give you the power that you have to then use to meet the loads so this is going to be a live digression into power system frequency the power system is integrated synchronous machine the motors that are running the generators that are trying to produce the electricity they’re all spinning at 3600 rpm they’re all at 60 Hertz and the the frequency of the AC sine wave is a direct measure of the state of the system from the standpoint of the balance of the load the electricity that’s being withdrawn from the system and the electricity is being injected into the system and so this analogy of using the water level in a container is very apt for exploding power to some frequency concepts if the inflow on this left-hand side is there a pointer did they give me here c’est la vie in Flour is equal to the outflow the frequency will be stable at 60 Hertz if the inflowing see if exceeds the outflow the frequency will rise if the outflow exceeds the inflow the frequency will lower and that’s a very simple mechanical it’s much more complicated than that but it’s a basic idea you know in and out or the reservoir changes levels so there are two ranges of frequency operations that are concerned to the power system during normal operations the system frequency is maintained through very explicit control actions within a very tight narrow range if this were a class I’d ask the question what why that’s the case and maybe we’ll come back to that in the question and answer session the concern that we have is that the power plants break and in transmission lines break randomly and in order to keep the lights on you have to be able to sort of restore frequency very quickly and so I’m going to be very focused on what happens when a large generator or a large transmission line is lost and frequency falls precipitously and you have a series of control actions which try to restore the frequency so the bounds of frequency are much wider frequencies gonna fall much further and there’s some very extreme control actions that you’re seeking to avoid operating well the first one being under frequency load shedding which means large blocks of customers you know entire regions of California are turned off in a last-ditch effort to try to restore the balance between generation and load or in the extreme when that fails the generators take themselves off to avoid mechanically damaging the turbines themselves and the philosophy there is is better to be able to restart

after the blackout than to destroy the machine trying to prevent the blackout and so what I’m going to be very much focused on here is these contingency situations where something bad has happened and you need to sort of restore frequency quickly in order to keep the lights on and the frequency response more broadly speaking refers to that process and it’s different both in the amount of frequency that you have to operate over and in the timescale that you have to exert those actions than normal frequency control which is a minute-by-minute type of balancing activity that goes on continuously you know explain those in the next slide here this is the anatomy of a frequency response event in the top I’m going to show frequencies evolution over time on the bottle I’m going to show the different control actions which stem and restore frequency over time and there’s two different timescales the first 30 seconds which is the critical time which frequency must be arrested or blackouts ensue and then this restorative period in the minute time frame when you’re kind of restoring frequency and trying to rebalance the system so what happens when there’s an immediate in so when a generator is lost there’s an immediate imbalance between the load generation balance you have less generation you have load frequency begins to fall the rate of decline is determined by the inertia of the power system which is a function of the total installed generation and load that is connected at any one time instant in time that so depending on how big your system is depending upon how much that imbalances that will determine the rate of decline that decline at this point can only be it arrested by the actions of what’s called primary frequency control these are autonomous distributed control actions that are exerted by generators responding directly to frequency through something called a frequency droop requires a generator both have some room to move not be loaded up to the top because there’s no place to push up and also have a governor which allows it to sort of increase its output in direct response and direct opposition to the amount of frequency deviation from the nominal value these actions must take place in the first ten seconds in order to arrest frequency before it’s declined to the point some of these more extreme actions that I show on the prior slide take place they will stabilize frequency of what we called an idea of frequency and the key here in the analysis that may be doing is to show that the adir does not go below we go back a slide the highest under frequency loadshedding trip what you want to do not want to get to this point so you want to arrest frequency up here and you can see immediately because it’s this opposition of forces here that where this nadir is formed depends on how much of this you get up and how fast and so this is the focus of what I’m going to be talking about once the nadir is arrested if you have more juice in the tank because it will actually stabilize at a higher value and at this point slower forms of frequency control some that are more familiar to you if you study the wind literature this is called AGC or regulation secondary this essentially dispatched a commands to generators to move up in response to the deviation in frequency these actions take 30 seconds to 10 minutes to be effective much too slow to arrest this free fall frequency that happens when you immediately lose generation so again what I’m trying to do is establish the premise see in the essential role that the primary frequency control provides interesting the drop in frequency in preventing blackouts once secondary frequency control as recently as allow the frequency to come back to the target value you have what’s called tertiary frequency control which is generally as being moved up and down to restore the reserves that provided secondary control and the reserves that provide a primary frequency control and basically the idea is you want to position yourself for the next unplanned contingency which could take you down so again it’s a sequential set of actions and primary frequency control for the purposes of this discussion is the most important of the three so this work we’ve developed a series of metrics to measure the adequacy of primary frequency control the first is kind of a lagging metric on primary frequency on frequency response current practice is to measure frequency response based on the amount of generation it’s lost the number of megawatts that are lost in the deviation in frequency from the starting point at a and the settling point at B the reason for that definition you know and this is

you’ll you’ll see immediately this is kind of a second-best measure because what I’m really worried about is whether this point is going to cause under frequency load shedding is because the traditional methods for monitoring the power system take a snapshot the system only four to ten seconds so you can’t even observe this with the current generation of monitoring technologies that’s changing we can talk about that later so this is the only observable what you had once frequency was restored and so that’s why the traditional measure was based on on the current generation of monitoring technologies that’s used to operate the grid in our report we proposed using a new metric based on an idea this is what you’re worried about that’s what you should be measuring with the new generation of syncro phasor measurement technologies we should be able to measure this more accurately in the future the second set of metrics that we are define were the amount of primary frequency response that’s provided by these sources of primary frequency control and what you can see is it’s a time varying metric about how much power is being injected at each time step and so you can see in order to arrest frequency depending upon how fast frequency is falling a function of the system inertia and the amount of power that the amount of power imbalance that will dictate directly how much primary frequency control you might need to arrest frequency at any given point and so in this report we outline a method for establishing the adequacy of primary frequency control we ask that you determine what is the contingency that you’re concerned about what’s the largest loss of generation event that you want to protect against you want to then decide what’s the highest setting of under frequency loadshedding that you want to avoid tripping and you wanted eternal how much margin you want to have between that highest shed point and the frequency nadya that you’re going to establish and that would define explicitly and directly your requirements for primary frequency in control both how fast you need to deliver it and how much you need to deliver and that’s the methodological contribution of the sort of set outline that process so now what I want to do is talk about how variable regeneration affects frequency response the first phenomena which is something that people very concerned about initially was that it will change the inertia of the power system it’ll change the rate at which frequency falls and the reason for that is that all modern wind turbines are electrically coupled to the power system through essentially a power electronic interface that pilot Roenick interface effectively decouples the inertia that the wind turbine has through its spinning blades and contribute to the inertia of the power system so it’s basically a zero inertia contribution and so who have a lot of wind it’s displacing the generation that would otherwise be serving that load and so you’re basically withdrawing a certain amount of inertia that would have been provided had rotating generators conventional generators even providing meeting that load and so your concern about loss of inertia which would mean that this this free fall would happen even faster and see in and and so you would need even more of this we find that to be a minor effect at the levels of wind that we’ve been studying you can envision a system which has you know extreme amounts of wind power or inertialess sources in which this will become a bigger problem it’s only this wasn’t going to be an issue what we saw well we were more concerned about is the second effect which is again the wind is displacing generation that otherwise would have served load and all the wind turbines and the incentives that are given to wind turbine operators is to run at max output that means you have no place to go if you have one these kind of frequency events in addition most wind turbines don’t have governor control that they can’t have it so what you might have as a situation where the introduction of a lot of wind would displace the conventional generators that had a governor that had Headroom whom you might have otherwise been relying on to provide you with this primary frequency control you might have lost some of what you were hoping was there this is a effect that we studied very very explicitly in our simulation studies another effect which I think we’ve brought to the fore through the work that we’ve done for the first time is that you’ll by displacing this generation and by replacing with wind yes I necessarily a one-for-one substitution because you know again the winds blowing where the people are not the generator notice over the people are and so to deliver the primary frequency control you have to rely on the existing network of the transmission system so you have to do actually a transmission system analysis to determine whether you can deliver the primary frequency

control you might have enough primary frequency control but if it’s in the wrong place you can pop one of these transmission lines trying to deliver it very quickly so there’s a deliverability analysis in addition to an adequacy analysis that has to be conducted the final effect that we examined which I think was very very underappreciated prior to our writing this report is that in normal operations in order to keep frequency here you’re relying on slower forms of frequency control you’re relying on these secondary frequency control actions okay and so this so if you’ve studied these wind integration studies they talk about how much more regulation do you need how much more load following do you need to the extent that you have under forecast your requirements for those regulation requirements to the extent that frequency deviates more than your secondary frequency control can correct for you begin being into these primary control resources automatically because it’s the inadequacies in secondary control are manifest in deviations in frequency if these deviations in frequency are greater than the secondary sources can handle then the primary frequency control resources are going to start getting activated to respond and try to restore frequency and the concern is here that if you’ve under forecast these secondary reckon dairy frequency control requirements that you may have nothing and you sub predating you start you know eating into your primary frequency control reserves you may not have enough primary frequency control left for when the really bad event happens which is a sudden loss of generation and so you’ve used up this precious resource which is the only way that you’re going to get out of these very extreme situations the other thing our study clarified is that maybe we’re looking very concern about the rapid ramping of wind and we try to show in our study as you know that is a very important operational consideration but it is not the same kind of a frequency response event as a lot of sudden loss of generation or a large transmission line you know loss of a generator or transmission that’s in a second that’s in an instant these ramping events minutes or hours to evolve I will talk about an interaction between that of those types of events and this concern about the secondary frequency control but it is not a contingency that is the same as the sudden loss of generation and we were able to clarify that with this work this is a funny picture this was in here this was the picture of civilians so what I’m going to now talk about is the simulation studies that we did to try and demonstrate some of these concepts in these simulation studies we will talk about both the effective moral win and changing system inertia and we’ll talk about the effect of winning potentially displacing primary frequency control reserves we will not talk so much about this effect and I’ll give you an example of how this effect happens that can’t be studied with the kinds of simulation tools that we used so what we did is we got the system models that are used by each of the interconnections to do these kinds of planning studies and we ran them through a number of worst-case scenarios so the worst case scenario for this displacement of primary frequency control is typically at night so what at night what nighttime means is that the system load is very low and in most of the u.s. that means when the wind is at its maximum so you should have very little bit left for the conventional generation to serve and that’s the same conventional generation that you have to rely on for primary frequency control so the concern is under these very low load conditions high wind conditions do you have enough reserve left with this remaining set of generation to arrest frequency in your worst contingency so it’s again you know these guys are the power system that kind of thing to worry you know that’s something I think what’s the worst thing that could happen to me so you come up with these kind of worst-case scenarios and if you can show you know you can survive then you feel a lot better about you know kind of running the power system in fact the rules are all set up around assuming the worst and demonstrating that you you’re going to do that able do it so we got these system models from these different entities the acronyms at the end are the the specific dynamic simulation modeling tools that these simulations are setup and these are the exact same things that the planners use now there’s something that I want to say about this which is it was very important for us to use the same tools the planners used to do this demonstration because the idea was to show them they can do this stuff themselves and not to look at a reduced form system or a simplified system the limitation of doing this is that these are very detailed models of quarries specify the operating characters of every generator and every transmission line and so we were limited in the

amount of wind levels we could study to what was already in those models because what we didn’t want to do was say you know study a 50% win scenario which requires to both hypothesize for that wind generation would be located and then even more controversy you know hypothesize whose transmission line would be built to serve to bring that generation to load there’s tremendous I discussed in my earlier slide this transmission plan question is a very political question right now and we didn’t want to become mired in the politics of transmission playing to demonstrate what it’s essentially an operational issue for the operation of a power system will automatically so that limited the amount of wind that we could study to the levels that are already assumed in these models so let’s talk what we found so this is a simulation finding from the West in which we made two assumptions we studied both a high reserve and a low reserve case the high reserve case refers to a situation where at night you know your load is low and the the rules of the Western interconnection require you to have a minimum amount of observe reserves online at all times and so that’s what the low reserve cases the fact is because power plants don’t turn on and off so easily many power plants are just sort of kept online but run at low load at night so they actually have lots of reserves so we have a high reserve case as well and these are the brackets around them and then we run the cases for different amounts of wind being turned on at different at different times and so this is for the sudden loss of what the Western interconnection considers there was contingency which is the loss of two big nuclear power plants at parallel very at the same time which is about 2,800 megawatts and so what this simulation the results shows you is that even under these low reserved cases on I forgot one thing in the West the highest trip point for under frequency lurching is fifty nine five so if you can say about fifty nine five it’s all good and so that’s what we show under any of these levels of wind on these very bad conditions of low reserves so that’s a good story you know the way the West is going to be okay with those levels of wind in these amounts of reserves and that translates about 3% of the interconnections electricity requirements in 2012 this is a little more complicated this is the high reserve case for the West this is the low reserve case for the West these are the plots of how much primary frequency control is being exerted by the governors in both cases I didn’t have them separate for each case this is under the low reserve case and this is under the high reserve case and what this shows you there’s a couple of things I want to point out here one the rate of frequency decline is about the same under either of these cases and what that means is that this changing amount of wind in terms of displacing system inertia is not at that big a deal at least over the ranges of wind that we looked at what they emphasize however is that this is under the high reserve case that leads to this result this pink ones are the low reserve case which leads to this result and I’m sorry that they’re they’re basically identical slide this is demonstrating it’s how much primary frequency control you exert determines where you can arrest frequency you desert more of primary frequency control frequencies arrested at a higher place and also when you have high reserves you have higher system inertia so the slope is a little bit a little bit flatter under the low reserves case less generation is on line the slope is a little bit steeper you have less reserves the nadir is formed later and at a lower point so what these flies are attempting to illustrate is that this primary frequency control that’s the story you got a primary frequency control it’s going to be all good you don’t have enough gonna be a problem this is the basic story of the entire study here’s what happened in Texas Texas is really interesting and for those of you are fans of demand response you’re gonna really love this one because Texas gets half of their primary frequency control from demand they have under frequency relays on loads that participate in their ancillary services market Texas actually has rules that limit the the interconnections purchase of these reserves to about half the requirement because you know load basically bid zero price they always get selected and so what you see here this is a very dramatic example is this is just a presentation both of the primary frequency control exhibited by the governors and by the effect of the load dropping immediately so what happens is this is a huge jolt of primary frequency

control and it’s really been instrumental in Texas and think about Texas Texas is you know like one-tenth the size of the eastern connection it’s about it’s less than one-third the size of the Western interconnection so it’s a small system comparatively speaking so the same amount of loss of load you know two big power plants that’s a person on a percentage basis a bigger portion that interconnection you know because the size is so much smaller the inertia is so much lower so when you look at these frequency plots in Texas it’s really steep and you can see that this load effect is an important reason why they’re able to keep the lights on in Texas because it’s a very rapid injection of primary frequency control to stem the very fast falling a decline in frequency that results when you lose a big generator so this is good news for the defense of demand response the evidence can be as good or better than governor’s in some cases this is a less happy story and this is a source of stuff look I’m doing now when we use the eastern interconnection model and we’ve try to replicate it a known event we got the red line said frequency of the Reston here this is what happened we had syncro phasor measurement devices in the field when that big event happened it was actually a historic event this is what they told us this is the recorded OBE so we couldn’t reproduce reality with the eastern interconnection model and a lot of my work in the last years has been to work with the eastern detection entities to try to fix their model so they can do these studies more with more confidence and be able to reproduce what they’ve actually seen so now I’m trying not to my recommendations one of the studies that we did was to look at the frequency response may that metric that I told you about the kind of current metric it’s in megawatts 4/10 of a Hertz it shows that the frequency response these interconnects has been declining over time and so the concern that’s led to a lot of look in the industry is to sort of understand the causes of that it’s sort of what has led to this behavior you know the models think you’re going to do this the observation is what actually happens and so I’ve been doing a lot of work supporting the interconnections to do more analysis of their frequency response tune up their models and in fact they are uncle it on to the next one so the next recommendation so this is you know you should study whether this is a problem for you industry is doing that so that’s a good news the second thing that we found is that the rules that reliability the reliability that are should be requiring you to have adequate primary frequency control are imprecise in their specification and these are these are the reliability reserves that you’re required to have regulating reserve these are the online reserves are here up these are offline these are ones that must be able to respond in 10 minutes or less these are what are called spinning reserve and non spinning reserve non-steam isms offline sping reserve is online and you have rules about how much of spinning reserve you must have and you have rules about how much operating reserve you must have which is this you have general principles about how much contingency reserve you have but none of the rules actually say how much primary frequency control you need to have and primary frequency control is going to be provided by any online resource that has Headroom and that has an operating governor so you could have other online reserves other reserves that are online that you’re drawing from or you could have the ones you called the spinning that you’re drawing from but interestingly you can have spinning reserve online that doesn’t have governors it doesn’t provide this response and that’s actually allowed under the rules so one of the things that’s happened since this report is Newark last December bounded a frequency reserve frequency response standard that will require adequate frequency response from all the interconnections and so on my colleagues and I have been doing a lot of works that do the technical analysis of these kinds of things to try to help them establish where that standard should be set Newark has voted on that in December it will go to their board of trustees this week if the Board of Trustees votes it out it will go to FERC for ratification into a public comment process and if FERC ratifies it in this spring then we’ll go into a field trial demonstrating that this preserve can actually be implemented and measure and compliance can be demonstrated and we’ll be supporting that activity as well so those are two very good things that have come from this study in terms of trying to bring up the awareness of frequency response

as a concern for reliability and to sort of look very closely at the rules that currently govern the provision of frequency response to try to get them fixed now that there may be a quarter frequency response those of you are fans of markets you could envision frequency response products being procured competitively and that will provide value to those who are providing them now and to those who seek to provide them with innovative technologies in the future so that’s the third recommendation which is you should expand the capabilities and so one of the easiest things to do is to look at the existing fleet of generation turns out one things we do is we surveyed all the generators in North America for Newark to find out you know do they think they have a governor is it operating how’s it operating and you find out you know all Jerry’s first we’ll have a governor but many cases that are just turned off and for many may cases they’re trying for reasons that no one can actually explain one of the people we work with does generate a certification he goes these plants to demonstrate that they can perform according to the Newark rules and he goes how come the governor is turned off and the operator says I don’t know what that was so they push the button the governor’s turn back on so you know the losses is not through some will of willful malice on the part of that generate someone it’s just an educational question there is a very important issue though which is that generators are often under contracts to put out a certain megawatt output and so you can have a secondary control loop which after the primary frequency control is exerted says oh you’ve deviated from your set point go back to the set point ie take away the primary frequency control you were contributing to help keep the lights on because you got to meet this contractual obligation to have constant output so that’s a that’s a bad thing and so we’re working on on rules to try to address that as well the other recommendation the capability demand response you see in Texas that half their demand response half of their primary frequency control comes from demand response and I work with a lot of folks that are law the labs and universities who were showing you can do these from any other all kinds of unused appliances anything that’s capable of something frequency and as Alex will tell you can sense it but your plug so there’s no reason why you can’t do it in your own home you could also expand the capabilities of the renewable generation itself there’s no reason and in fact all modern wind turbines have the option to have governor connect of controls installed on them you’ve got to address the financial disincentive however which is having frequency control on a generator on a wind turbine means feathering back the output from full output that means foregoing revenue so less you can compensate you know you’re not knows how to give you this stuff for free so you gotta figure out how much you need you got to put a value on it and you got to sort of go out and try to get it and then fire there’s a lot of advanced technologies that could provide this kind of service as well for we we feel in particular and this is that there’s a limitations these simulation studies and if the real issue with a lot of this variable degeneration is short-term forecasting and prediction and operating procedures that posterior system to be able to respond to these unplanned wind events so this is an example of what happened in air cot in one one period so what you have here is the amount of wind falling pretty fast three times that day then you have what’s going on with system frequency here if we can see as falling as the wind output is going down because it’s trying to get generation there you’re trying to meet most of that of that fallen frequency with secondary control okay you got to this point here at this point of the day where you’ve run out of secondary control so you started using primary control this is called the rapid response reserve this is what they call primary frequency control in Texas they actually dispatched it they just said turn it on we got to stop Requa so now what I’m not showing is that they’re bringing up other generators that are also contributed and that restored the frequency so you’re okay fine mill a day and so secondary goes back down you can dispatch this is the tertiary stuff it’s coming back on so you can bring these guys back down to this this even point yet another one ramp event here it looks like you were fine you know you didn’t even use a secondary you had enough other resources online that you could generate the powder it’s then you add a third one again here and again you had to dip into this primary frequency so this is that that fourth effect that I was trying to describe to you earlier and this is a real-life demonstration of this concern about if you don’t have enough other stuff you’re going to start eating into the primary frequency control and that’s why it’s a very important concern to have adequate tools to predict these kinds of wind ramps so you could have enough secondary frequency control or load following so you can always keep this reserve of

primary control control available when he’s bad if you had a loss of generation event at that time and that will cause better operating tools better training better forecast etc if recommendation is that these are these are plots of the frequency response of all the interconnections this one is the East it’s not for this Auggie’s of different events this is the West this is Texas and here you can see how fast frequency falls in Texas compared to the East which is so much bigger which is you know frequency control is a feature of all interconnections and as the composition of the fleet changes it’s appropriate to begin examining whether you have enough frequency response and so I’ll give another example when this study was written there was a lot of interest in more nuclear power nuclear power is typically dispatched full-out and by agreements before the NRC does not provide frequency control so here’s another source that’s on the system that’s not contributing frequency control or frequency response for that matter potentially displacing the other sources that you are relying on the same kind of procedures can be used to study that same technique same methodology and that’s the kind of proactive approach that we advocate which i think is starting to take hold a number of the interconnections there have been a number of studies done on frequency response of the interconnections of individual entities within the interconnections inspired by the work that we did for FERC so that’s kind of what I’m going to end I’m going to just tell you what’s in these reports because this is where the really really interesting stuff is this first paper was written by the person these models here these dynamics simulation models pslf pssc mr. Underhill wrote those models for GE twice and for Siemens PTI once and he wrote a primer on power and frequency control concepts that is just a wonderful wonderful text book if you want to learn more about these control concepts I’m trying to get him to turn it into a book the second one is work that my colleagues did by using tools that my consortium has developed and that are you by nerf to monitor frequency in the internet connections to actually do the analysis that led to these results here the third paper is written by the fellow who is the author of the control performance standards that are used by Nurik to determine how much frequency control each interconnection each balancing authority must have if you know the Nurik operating procedures or operating rules now these stands are called the control performance standard one control performance standard – cps one cps – as well as the disturbance control standard he was involved the original committees that developed those and he basically talks about you know in the old days everybody had a governor everybody had it enabled we never had to worry about frequency response as a result of restructuring we’ve had to worry about it because no one wants to provide anything for free anymore and in fact because our rules you know we assumed everybody had it we never had to write a standard about frequency response because we just always assumed that we would have enough now we need to be more explicit he was very involved in the drafting of this new rule that has just been voted out this third fourth piece or the actual simulation studies that we conducted that I showed you some examples from and the fourth one the last one is a very interesting paper by my colleague up at the lab she’s a statistical physicist who did sort of a bottom-up analysis of the variability of wind and load and what she showed is that wind power very variations in wind power follow a power law rather than a Gaussian law in terms of the distribution and all the studies that have been done to date about operating under lots of what have assumed a Gaussian distribution which means that they have under forecast extreme events which are exactly events that were concerned about for these power control situations she’s going to continue that work and we’re in the process of talking with a lot of system operators about trying to incorporate a work into this short term load forecasting activities these are all downloadable in the public domain I encourage you to go to them now with that I will conclude and take some questions Dave Watson Lawrence Berkeley Lab a great talk Joe thank you with increases in use of gas fire turbines and lower gas prices how do you think that’ll change the equation and our newer combined cycle gas turbines more responsive in in other words able to ramp up and down more quickly and can

you modify existing gas turbines so that they’re more responsive as well one of the great things about my project was working with John Underhill cuz he worked with GE Power Systems for many many years and he was very blunt with me you pay them they’ll do it you know there is now a whole class of GE combined cycle turbines which advertises their fast ramping capability you know it’s just you tell the manufacturer what you want you tell them you’re gonna pay them for it they can make these things do whatever you want them to do so there’s not a technological limitation in terms of the turbine technologies to be able to provide this kind of frequency control but your feelin’s might be harder but let me just say that in terms of primary frequency control all generators have governor’s on them whether they’re are enabled whether you’ve run them in a way that allow them to provide this program that’s a discretion of the operator not but it is every generator that you can buy you can’t buy without it or that way okay Oh what what is the percentage of wind power there’s a percentage of the total generated power in the u.s. desert just roughly what what what is the percentage I used to have those numbers at the top of my head you know I think it’s still on the order of two or three percent in terms of total generation you know as a fraction of total energy generated installed capacity is a little bit higher so I would say you know maybe 7 or 8% of total installed capacity because you know the wind doesn’t blow all the time so usually assume a capacity factor of about 35 so that’s a rough translation they’re much higher in Texas I should I just said that you know the reason why we had to do three different simulation studies is because each of these electrical interconnections is electrically asynchronous with one another they are synchronous within the interconnection but they are air sync risk with respect to one another and that’s why we did three studies one of Texas one of the west and one of the east because within the East they’re all one machine within Texas or the air cut part of Texas they’re all one machine and within the West it’s one machine Joe Ron great great talk could you talk a little bit about the use of demand response and the different modes of regulation that you were talking about and and maybe related to the success that they’ve had in Texas and try to classify what are they using to get the good responses well you know and from my perspective demand response you know particularly managed as an aggregation of demand responders can provide any of these frequency control services you know it can provide load following it can provide regulation and it can provide primary frequency control and there’s a lot of demonstrations going on around the country and in fact in many of the markets Jason could tell us about this may the markets demand is actually participating in providing regulation services so I don’t see you know there are barriers to entry that we’re doing a lot of research on to try to address a fundamental part of it is trying to reveal the value of demand response so until there was a standard for primary frequency control there won’t be a market for it until as a market for it people get paid for it and so that will determine that the rate of development of primary frequency will control from demand response in Texas they have a market for it they allow you to do it in other markets it’s not not quite as mature Praveen I’m sorry alex is yes really interesting have you thought at all about frequency excursions caused by abrupt loss of large load as opposed to abrupt loss of generation in particular and I’m thinking of the earthquake warning systems in Japan causing the loss of hundreds of mega washer in the train systems and now we’re talking about installing that in California are we gonna I don’t know what happens in the other direction if you suddenly lose 500 megawatts of load well two things typically the kinds of low loss events that we’re talking about are small compared to the large generator loss sense that people are concerned about and from the what I’ve seen the loss of generation is much more frequent and a much more routine occurrence and that’s why you know there already have these

spinning reserve rules but there are over frequency relays certainly for the keep the generators can going too fast so you could I pop the size a situation where if there’s a common mode failure of a lot of load you know and again this is all about sort of like how much insurance do you want to buy you know what are the you know are we playing for the earthquake all the time as opposed to you know these generators break every week and if something goes wrong you know that we built a know difficult a notification system for Nurik that says when these frequency events are happening you know and it sends out a sends out a pager alarm talk you know the selected list of people like in one of those every other day you know now they’re not big events but you know these things you know there’s hundreds of generators out there there’s something breaking all the time so it’s not unusual you got to give the to a Praveen to next after after you okay okay sorry about that go ahead thank you so you mentioned that currently the renewable energy machines on the on the cria don’t really participate in the frequency response so far they’re generating at 100% and there are not a lot of economic incentives that they would participate in these frequency response measures other already any kind of regulatory reforms would say that they might not operate it at 100% but ninety-five percent or so to move up and like who’s regulating that is that physicists which has the greatest frequency response control problems because of the smallness of their system and the largest of the events compared to the size of the system they have a requirement that every winter and must have an operable governor well at this point and so they in principle could participate in providing their rapid reserves they’ll mandate that they do so because they set a target they have a market and they procure it but they do require every every wind generator to have a governor operable governor so they could potentially participate yeah so that those rules are set on the interconnection level and not at the national level so well they could translate into rules at Newark they haven’t yet they would typically be done start regionally and then move up if there was determined to be appropriate Praveen is there anything magical about fifty nine point five versus fifty nine point two I mean I can’t imagine how ie being able to meet your standards that’s a great question I was hoping you would ask something like that because the reason for tight frequency control in interconnected power systems is explicitly for commerce because if you’re deviating from that 60 cycles that means somebody’s generating more in somebody’s generating less and the idea of tie line control between interconnected systems was you each keep your nose clean everybody matches their own generation with their own load I mean they can buy from somebody else but as long as you could hold it at 60 that meant that you were imbalanced the moment you become out abouts somebody is serving your load or you’re serving somebody else’s load and when money becomes involved you want to be compensated fairly so that’s why all interconnected processors have this very type frequency control requirements it turns out in terms of equipment damage you can go a lot lower than 59 five before these turbines get damaged and if you want to pay the manufacturer more money I’m sure you can go even further down all Island power systems which are very small inertia right because there’s small isolated systems they have wild frequency excursions you can see the data for Ireland which is connected to the rest of Great Britain with us with a DC line which effectively decouples it you can look at Hawaii and you can just see that these curves go all over the place and that’s really they don’t care you know a clock I mean most people use digital clocks now anyway so they don’t about the clocks going faster slow and so in principle now of course there’s all these systems that have been set up around this type frequency control regime like the under frequency load shedding relays but in PI excuse me oh sure they used to do that they’re doing more and more of that you know but you know it’s called inadvertent exchange and you as opposed to a lot of the Newark rules are set up around you got to pay back in kind and as long as you keep your nose clean on average I mean that’s what the control performance standards are all about on average showing that you’re in balance and not leaning on your neighbor but it’s impossible to hold 60 Hertz all the time and it’s impossible not to be leaning on your neighbor some of the time that’s just the nature of the beast the question is how much and for how long

great question okay I think last question all right Merlin thank you Joe you found in this study that the dilution of inertia in the systems you looked at didn’t have a scary impact but the way you stated it sort of implies there may be a place where it becomes so diluted that it will protect you have to have an impact another unknown in this it seems to me is the change in the customer load traditionally an induction motor type load – more and more power electronics look particularly as more and more photovoltaics get on there and so I don’t know what then there’s a couple of comments there so yeah that’s what I want to know what where are we going is anyone studying it and I any projections oh well you’re studying okay so two things here there are inertialess power systems another project what’s called the certs micro grid and it’s all power electronics and you you manage frequency directly you know it’s a drupe that you program in and that’s how he balanced load and generation so you know you can envision these these situations now load does have a frequency response component the motor slows down when you go below 60 cycles and so what I don’t show in here is in addition to primary frequency control coming out of the governor’s there’s about one-sixth the amount coming out of motors slowing down and and but if you have a motor that has a variable frequency drive on it you could actually see the reverse effect some of those motors are designed for constant speed so they’re going to take more power when you want them to be actually slowing down so there is a different interaction with new types of load particularly those with power electronics and we’re doing a lot of research both on it from not so much from a frequency standpoint but from a voltage standpoint I guess that’s it thanks so much for coming hope the film is good