 Well, hello, and good afternoon, good evening, or good morning, depending on where you're joining us from today. Welcome to Engineering for Change, or E4C for short. We were pleased to bring you the latest in our 2017 webinar series on the topic of innovation in microgrids. My name is Yana Aranda, and I'm the president of Engineering for Change. The webinar you're participating in today is part of E4C's professional development offerings. Information on upcoming webinars in the series, as well as archive videos of past presentations, can be found on the E4C Webinar's webpage, as well as on our YouTube channel. You can see both of these URLs listed on the slide here. If you have any questions, comments, and recommendations for future topics and speakers, please contact the E4C Webinar series team at webinarsandengineeringforchange.org. If you're following us on Twitter today, I encourage you to join us on the conversation with our dedicated hashtag, hashtag E4C Webinars. Before we move on to our moderator and presenters, I'd like to tell you a bit about Engineering for Change. E4C is a knowledge organization and global community of over one million engineers, designers, development practitioners, and social scientists who are leveraging technology to solve quality of life challenges faced by underserved communities. Some of these challenges may include access to clean water and sanitation, sustainable energy, improved agriculture, access to information and internet, and more. We invite you to become a member. E4C membership is free and provides access to news, information on hundreds of essential technologies in our Solutions Library, professional development resources, and information on opportunities such as jobs, conferences, funding calls, and fellowship. Members enjoy a unique user experience based on their site behavior and engagement. Essentially, the more you interact with the E4C site, the better we will be able to serve you resources aligned to your interests. For more information, please see our website to sign up and explore. Now a few housekeeping edits before we get started. Let's practice using the WebEx platform by sharing where in the world you are joining us today. In the chat window, which is located at the bottom right of your screen, please type in your location. Note if the chat is not open on your screen, you can just click the chat icon on the top right corner. And I'm going ahead and type in my location. I'm dialing in today from Brooklyn, New York. And let's see where folks are from. And I see we have participants from Toronto, participants from Chicago. I see some folks coming in from India. We have a few from the United States. Thank you so much for joining us. We're so pleased that you've made it today. Please know you can use this window to share your remarks during the webinar. And if you have any technical questions, just send a private chat to the Engineering for Change admin. During the webinar, please use the Q&A window, which is located below the chat, to type in your questions for the presenter. Again, if you don't see that window, you can just click the Q&A icon on the top right corner of WebEx. If you're listening to the audio broadcast and encounter any trouble, try hitting stop and then start. You may also want to try opening up WebEx in a different browser. See first-you webinars qualify engineers for one professional development hour. To request your PDH, please follow the instructions on the top of the E4C professional development page after this presentation. The URL is listed on the slide. And a special note about today's webinar. We will actually end at 15 minutes past the hour, or 12.15 Eastern Standard Time, to ensure that we have enough time for Q&A with our incredible panel. And with that, I'd like to take a moment to introduce you to our moderator for today. Frank Berg is the VP of Grid Engineering for Segora International, with experience in renewable energy in a variety of contexts from community-driven systems to utility-scale power plants. He's an instructor for villagers, teaching web-based courses on appropriate technology and community-based development across our state university. He's been an active leader within Engineers Without Borders USA since 2005. And Frank is also a contributing editor at E4C, has been with us for over three years. You can follow Frank on Twitter. His handle is right there, but also on the E4C platform. And with that, I'd like to turn it over to Frank to introduce our panelists. Well, thanks, Yana. And welcome to our participants from all over the world. As Yana mentioned, my name is Frank Berg. I'm a contributing editor with Engineering for Change. And you can find some more of my thoughts on the blog. I want to introduce today's presenters, Henry Louie, President and Co-founder of Kilowatt for Humanity, Omar Ghani, CEO of Kilowatt Labs, and Jay Taneja from Assistant Professor of Electrical and Computer Engineering at the University of Massachusetts, Amherst. In order to start the conversation today, I'd like to just kind of introduce the concept of electrical microgrids in the context of energy access and rural electrification. First of all, Sigoura International, my employer, is a company that builds micro-utility business models by building the grid from scratch in places that have never had power. The photo here is one of our technicians in the grid that we've built in Northwest Haiti. We currently provide electricity and keep the lights on for about 10,000 customers in three towns in Northwest Haiti growing every day. In order to set some of the context for this discussion, I'd like to point out that there's a high cost to darkness. People without access to reliable electricity pay far more per unit of energy than folks with reliable electric connections. As you can see on the left half of the slide, folks without grid access pay upwards of $10 per kilowatt hour for lighting if you do some math from dollars per watt and watts per lumens and $20 to $50 for cell phone charging when they're charging on the street at a vendor or traveling to the nearest electric grid or generator. On the right side of this slide, you have the average rate of electricity in all of the countries in the European Union as well as the average price in the United States as well as the average price for the national utility in Haiti. As you can see, it's 100 times lower for customers connected to a conventional grid than customers who are using subsistence electricity to which you're self-basic. The difference is $10 U.S. versus $0.10 of a U.S. dollar roughly from one to the other. So the punchline here is that poverty is expensive. In fact, 100 times more expensive than conventional grid connections to existing grids. So oftentimes folks will say, well, the problem with microgrids is we can't afford them when the reality is that our customers are actually paying 100 times more for what little electricity or energy that they do consume. So there is an ability and willingness to pay for some electrical services. The next concern that I often hear is that basically these projects are not financeable or that the investment would be better spent elsewhere. And to this, I'd just like to share some experience from the context of utility companies operating in Sub-Saharan Africa. Looking at the slide, the chart in the center shows the capital expenditures in green and the operating expenditures in purple for each of the 39 countries in Sub-Saharan Africa for their electrical utility. In many cases, these utilities are nationally owned, but in some context, there's multiple privately owned utilities in a country. The red dot in each bar represents the cash collected as a percentage of total expenditures. And you'll see only two out of 39 countries are able to cover both capital and operating expenditures with the cash they collect. So what that means is that the business as usual in Sub-Saharan Africa electric utilities is insolvent. The costs are either subsidized or not fully reflected in the prices that customers pay, or energy theft has resulted in the inability to recover the revenue that it takes to run these businesses. So it's not that microgrids are not financially viable. It's that electricity as practiced in developing nations is not financially viable. And I would say that there are plenty, this chart illustrates there are plenty of room for improvement in terms of the way that we build, operate, maintain, and collect payments for electrical services in the developing world. The picture there, Munoz, Soif, Courant is the unofficial motto of Segura Haiti. That's a painting on the wall of our office in Molten Nicolae, Haiti. And it means, in Haitian Creole, it means the people are thirsty for power. And it's true that people in developing countries need modern electricity access for many productive uses of power, from irrigation to refrigeration to water purification. All many, many different aspects of everyday life, including healthcare and education, are much more viable with reliable service to electricity. But these types of business models and business as usual is not getting us there. The cost for connection of grid extension is not reaching these customers as quickly as they need the power. And individual solutions for a variety of reasons don't meet the needs of productive use. So hence the interest here in microclips, the one being built in the background of the picture on the right. Again, that's from our system in Haiti. So one of the questions that our panelists will be grappling with today is how do we measure the impact of electrification in this context. And so here's the slide of a couple of the metrics and frameworks that have been put forward to analyze and compare these different systems. So on the left, you have tiered access to electricity structure from ESMAP. And it basically ranks different. It helps to quantify different levels of electrification from tier zero, which is essentially no electrical infrastructure, to tier one, tier two, basically solar lanterns. Basically a flashlight with a solar panel on the back of it. And tier three and four solar home systems and tier five, or tiers four and five microgrid systems with varying degrees of reliability, uptime and energy access. This is a framework that's been put forward basically to compare different levels of electrification. Certainly there's a different price point for a solar home system versus a full utility scale grid. But also there's different levels of use. And on the right is the quality assurance framework for mini-grids. It's been put forward by the National Renovable Energy Lab here in the U.S. It basically takes some of what the ESMAP standards we're trying to accomplish, but puts it into more technical terminology using more standard metric from the utility industry to measure reliability standards and kind of normalize across the different metrics how we can compare these from one to another. So that was just my quick words of introduction here to kind of set the tone for the importance of microgrids in developing countries and the discussion we're going to have today. So without further ado, I'll get out of the way and I'll pass the mic over to our first panelist, Dr. Henry Louie, the founder of Kilowatt for Humanity and professor of engineering at Seattle University. Yeah, hello, and good morning, everyone. I hope you can hear me. Frank, could you maybe just check that you can hear me? Yeah, I got you. Loud and clear. Okay, thank you so much. So yeah, good morning, good afternoon, good evening. My name is Henry Louie, and as our illustrious moderator, Frank, said I'm a professor at Seattle University and I am a practitioner involved in the off-grid electrification space through my nonprofit organization called Kilowatt for Humanity. We work in rural areas primarily in Zambia, but also Kenya and a few other places to provide electricity access or an energy kiosk model as well as sustainable business opportunities. I've spent quite a bit of time in Zambia. I was there for a year continuously as part of the Fulbright program where I worked in electrification and electricity access in general. Let's go to the next slide, please. So when we think about those tiers that Frank had mentioned, the different levels of access that have been defined as MAP and adopted by a few other organizations, you can really map some technologies to that level, to each level, each tier of electricity access. And I would say that sort of on the low end of it would be solar lanterns by various providers. There's some estimates that there's something like 40 million of these units that have been sold over the last couple of years and they provide a very modest amount of electricity access, maybe one or two LED lights and some cell phone charging. As we move up this curve, you see that the systems become more capable and sophisticated. You might have improvised microgrids or what you might call un-engineered microgrids that you would see in someone's house. Maybe they run a little wire to their neighbor and it's powered off an automotive battery or something like that. Higher quality systems might be called filler home systems. These are usually well-engineered products. They might be costing hundreds of dollars and they might be capable of providing lights, radios, even television access. Moving up further, you might have energy kiosks. These would serve a small village not necessarily through direct connection, but it would serve as an energy access point where there might be some refrigeration. There might be a place for people to recharge their solar batteries, car batteries or cell phones and really there's some productive uses that can be associated with that. Energy kiosks are typically a few kilowatt range. As we move up further, we get into the mini-grid or microgrid space and really the distinction between a mini and a microgrid isn't really clear. So I'm going to use it sort of generically here and call them mini-grids, but these would typically be maybe five kilowatts up to a few hundred kilowatts in size and they serve wired connections to lots and lots of customers, dozens to hundreds of customers. Most of it at the top would be the grid and the caveat here would be that in some cases and in fact a lot of cases, a well engineered mini-grid will outperform the existing grid, the national grid in terms of reliability. So those two might overlap if you look at the quality of access. So let's move on to the next slide here please. One of the things that we do in my organization, we don't own any of our systems, rather we have a nonprofit organization in the local community that actually ends up owning it and operating it. And I think some advice I would give to anyone involved in this off-grid electrification space would be that you have to pay attention to the economics. It sounds obvious, but anything that you do, any panel that you install or battery that you install, it's going to fail. So you need to be thinking about how you're going to pay to replace it, how you're going to pay your operating costs if you have salary. So here are just some images of ways that we've seen energy be used in productive ways from cell phone charging to making ice, preserving fish. At the bottom there you'll see wired distribution to homes and you can charge as Frank had mentioned, varying tariffs that will help you recoup your costs. We found that selling cold sodas has been very popular, generally with a very high profit margin when partners end up doing that. And then there's some more creative uses. What you see on the bottom right of the screen is like a movie night that they show. So they have a projector, they show movies on that white sheet that you see there, and they charge a small fee for people that want to come and watch the show. So you always need to be thinking about the economics. So let's move on to the next slide. Thank you. So this is going to be an animated slide. What we actually do, mini-grid designs, is the two main inputs really are some sort of estimation of your demand and some sort of estimation of your resource. And this is true whether or not it's solar or wind or even biomass or hydro, you need to know your input energy source. You take your demand and you take your solar resource in this example, and you might apply computer simulation or you might apply an IEEE or an IEC standard or maybe your organization has some sort of rule of thumb standard that you apply. And from all of this, you would output, please advance the animation for me. Thank you. You end up getting some sense of the sizes of your components and costs. So if you use, for example, Homer, at the end of the day, when you're done doing your analysis, you'll have the approximate sizes of the major equipment, the inverters, the batteries, solar panels, et cetera. And that's how we do things. Let's do one more animation for me, please. So usually getting the resource input is easy if it's solar. You can just consult a database. It's a bit more tricky, for example, if you have a wind-powered system, but you would set up a MET tower or you would try to find some historical data to get an idea of your resource. So it's fairly easy to get hard numbers on your energy resource. Let's do one more animation, please. Estimating the load, however, is very tricky. Really the state of practice that many organizations use is to do surveys where you go and you talk to the potential customers and you ask some sort of appliances they might want to own and you figure out how many hours a day they might want to use it and you look at the power rating of different appliances and then you sort of come up with an estimate of their energy use and then you design your system around that, as you can see. Now, there's a lot of problems with this approach. People who have never had access to electricity, I think, are very poor at estimating what amount of electricity they might actually use. It's aspirational. It's very predictive and speculative. In addition, if there's a tariff associated with the electricity, it's really hard to figure out that many elasticity looks around the different pricing points. I think this is an area that is prone to error. Let's move on to the next slide. Thank you. There's actually some very meaningful consequences of this error. If you happen to overpredict, meaning you predict that each customer is going to use more than they actually do, essentially you've oversized your system and most of the components, they're cost scale approximately linearly and some of the estimates put in about the penalty for oversizing being somewhere around $6 per watt hour of error and that adds up considerably. What this means is that for a given amount of budget or investment, you can only install fewer systems. So if you were off by 100% in your error, you were only able to install about half as many systems and that's a real challenge for companies and organizations trying to make good use of their dollars. As a consequence, you need to really charge more money. You have to take that capital expense and spread it over fewer watt hours sold. On the other hand, if you tend to underpredict, so you predict less than what is actually needed in terms of energy use, well, then your system is smaller. You spend less money on it, but the reliability probably isn't where you want it to be and this can drive premature failure because batteries are cycled deeper and so forth and you can actually lose some customers. You can see some customers' defects from your mini-grid if you don't price it. So some of the research that we've done, please go to the next slide, is to look at just how bad this survey-based error is and this is some work that we did with Vulcan Inc. on some of their mini-grids in Kenya and you can see this bar chart shows the results of that survey, the predicted amount of energy and the actual. This is for eight different mini-grids in Kenya and you can see that there's a strong bias towards over-predicting consumption and in fact, if you show the next animation for me, you'll see that the total predicted, if we look at this as a portfolio of mini-grids, would have been about 72 kilowatt hours per day, but the actual was much lower. It was only 17 kilowatt hours per day. So the capital expenses were bloated and the grids ended up being much larger than they needed to be. To give you an idea, the average consumption ended up being somewhere around 100 watt hours a day and I believe this sort of estimate agrees with what one of my co-panelists will be talking about a little bit later. Now as we looked at this data, and the next animation please would be helpful, that the actual uses were fairly similar. So there's one grid that sort of over-achieved in their consumption, but the rest were fairly close to one another and that led us to conclude that actually the better way of predicting mini-grid consumption is to actually just take a data-driven approach and I'll show you what that looks like on the next slide. So here on this data-driven approach, what you do is you simply consult the database and you basically compute the average person or per household consumption and simply use that to predict the future mini-grid consumption. And when we applied this approach and do the next animation for me please, we were able to reduce the error. So I think this really opened our eyes to the possibility of what data can do and if these results are known and we start sharing our experiences as an industry of what we think consumptions are, then I think we can really reduce our costs a bit. And there's actually a lot of other things you can do if you have access to data. If you go to the next slide please. Now this is just one example from one of our systems. Here we can see the solar power, the load, and battery voltage for a week-long period. This is for energy kiosk and silly baba. And if you show the first animation, please. Yeah, you see that if we look at the battery voltage, there's a certain level where it's the battery we can conclude is full or nearly full. And it's somewhere right around 28 volts for this system. Keep going with the animation, please. And then let's go ahead and keep going with the next two animations, yeah. And then at the lower end, there's a range in which the low voltage disconnect will happen. And generally speaking, you do not want this to happen because it means you're cycling your batteries. But this can happen depending on how you site your system. And in this case, we see that we do drop below the low voltage disconnect. And it really has to do with the cloudy days that you see in the solar power time series. So I have a call out in the next animation, please. So several cloudy days in a row, as well as a little bit higher than average demand, has caused the system to lose its load. And you can see in the next animation I point out where the inverter load basically goes to zero during those times. So having this data, we can diagnose the problem. We can look at what caused these power outages, which is a lot easier to support than just getting a phone call in the middle of the night saying the power went out, what happened. We don't have to send somebody on site to troubleshoot it. We know that it's simply an imbalance of energy in and energy out. And you can see towards the end of the week after we've asked people to try to limit their evening load that the battery is able to become fully charged again. So let's go to the next slide, please. So I think data acquisition is a must-have now in these mini-grids. I think a lot of practitioners are using it. And it helps with the design. It helps with de-risking investments so we can sort of unlock finance for mini-grids so we can show how much our grid's produced. Then it should be able to, you know, more easily attract investment. Operational decision-making, knowing when to shed load, for example, is something that can be done if you have these data acquisition systems. For my organization, we're a nonprofit. It's very useful for us to be transparent. And they, to donors and funders, you know, our grids have produced X amount of megawatt hours. Here's the reliability and so forth. And of course, it can be used for research and education. So let's move on to the next slide here. This is just my ways to reach out to me. You'll notice that there is a website, a link there, www.kw4h.org, where you can access our real-time data. And I'll show you what that looks like briefly on the next slide before I end here. Thank you. So we have a data dashboard where you can look at how our different grids are performing. Sometimes there's communication problems and you'll see no data for some period of time. But you can always browse back in time and see how our grids have been performing. If you're interested, maybe you're a researcher or maybe your organization has some interest in wanting to have the minimally data that we store, you can send me an email as you see there at the bottom and I'm happy to share the data with you. In addition, I'm part of the IEEE, the Power and Energy Society Working Group where we're creating a data archive for this type of information and you see the link there. And you can contact me via email if maybe your organization has some data that you want to share with us. So just one last slide here for the references that you might find interesting if you found part of my talk interesting here. And with that, I'll end my 15 minutes. And I guess, Frank, it's back to you. Thanks, Dr. Louis. Great job on the presentation. It's really good for our audience to hear some direct field experience and kind of what this has been like in your applications in Zambia. So with that, I'm going to pass the mic over to our next panelist, Jay Taneja, who is an assistant professor of electrical and computer engineering at the University of Massachusetts Amherst. So Jay, take it away. Great. Thank you very much, Frank. And I appreciate your content as well as Henry's content. I think they lead in very well to my presentation. So good morning, good evening, good afternoon to everybody on the webinar. Happy to discuss this with you today. So I want to talk today about that measurement problem that Henry so very well described. And this is essentially work that I want to share at work that I've done in concert with some students and some teams at Columbia University, as well as our partners at Kenya Power, looking specifically at measuring the growth and electricity consumption among grid and many grid customers in Sub-Saharan Africa. So this work, just background-wise, we see two very different pictures of electrification across the continent. If we're looking at, on a country basis, the map on the left here shows the electrification in urban areas. And generally urban areas across the continent tend to have somewhat better electrification, you know, 70%, but it varies from lower regions in the 20s and 30% all the way up. And then in rural areas, you have a very different story. You often have very low electrification. Across the continent, the average is about 5%. And why is this? Well, we know that the continent is quite large and that means you have a very sparse grid. So you can't actually reach all areas very well with a single interconnected grid. But the couple of factors I really want to point out here are that you have very high grid connection costs. These costs can be anywhere from $100 to $1,000 for connecting individual customers to the grid. And in many cases, that is beyond the reach of the consumer that actually wants that electricity connection. In places where those connections already exist, you still have a challenge where demand for electricity often outstrips supply. And so this results in frequent grid outages and you also have older equipment that can result in unplanned grid outages. So when you're looking at these two situations, urban and rural, you have kind of a key challenge in each one that we're highlighting. One is reliability in urban context. And in rural context, the question is really how you can provide low-cost access that fits the needs of these consumers. Next slide, please. So the big question, how much will these new electricity consumers use once they get connected and how will that consumption develop over time? So Professor Louie talked about the challenges in making those estimations. And I would say that it's a challenge for anybody to predict how much electricity they're going to use. In general, this isn't a quantity that we're used to how we can quantify it. It's hard to say what one kilowatt hour actually is. So the challenge, and we take this data-driven approach that Henry so well described, and so what we did is if you can advance the next slide, we worked with a data set from Kenya Power that looked at 160,000 utility customers. And why are we looking at utility customers and why are we looking at customers that are already on the grid? Well, these customers are our greatest guide to understand how new customers, when they get connected, will use electricity. And we took these customers and we built a classification algorithm that's able to identify which customers are in urban areas, which customers are in peri-urban areas, which customers are in rural areas. And really, when we're thinking about energy access, we really have to focus on these rural customers. These are customers that are as similar as possible to the next set of customers that are going to be connected. Most areas on the continent, as we notice, that are unconnected are rural areas. This is where electricity access is quite low. And by focusing on these rural customers, we want to understand what the electricity access and consumption patterns will be like for the next set of customers that are coming online. Next slide, please. So first looking at consumption among all customers. This is for the entire data set of $160,000. We actually have data over a number of years. And so this shows that for the first 10 years that a customer has had electricity, the red line here represents the median. So this is your typical customer and the dotted lines above and below the median represent your third quartile and your first quartile. And so this really shows this distribution. When a customer starts out, they generally start with very low consumption. We're talking three to five kilowatt hours per month, as you see very fast growth in the initial year. And then the growth slows down, but it continues all the way through 10 years. And this is as long as our data set can tell us, but we see kind of very good and stable growth among customers who have an electricity connection on the grid. Next slide, please. So honing in a little bit more, the story isn't quite as rosy. So this is for rural customers only. So looking only at those customers that represent our set of customers we're interested in for electricity access. And in particular, we're looking at customers who received their electricity connections in different years. So if a customer received their electricity connection in 2009, we have something like six to seven years of data for them given our data set. And that's that red line you see at the very top. And so for those customers, we see that their consumption peaks right at the beginning and then starts to level out. In fact, they reach their peak at around three years of having an electricity connection. And that peak is in the 30 to 35 kilowatt hour per month range. And so as we look at future years, we look at 2010, 2011 and so on, we have fewer data for each group of customers. So that's why the lines are shorter. But we also notice two very important changes in these customer bases. One, that the customers are peaking at lower levels. So the height of where their consumption goes is progressively lower as the customer has had, as the newer customers are coming online. And they're peaking earlier in their experience with electricity. So customers are peaking after 12 months instead of 36 months. This has something to do with essentially which customers are being connected as the grid grows. And we can see how this pattern is progressing that as we get to newer and newer customers, we're reaching the edges of our current grid and we're connecting generally lower income customers who have fewer appliances and consume less electricity. Could we advance the slide? Please, thank you. So I want to also point out what the cost of this electricity is. So this shows the tariff that these customers are paying. And in many countries, as Frank rightly pointed out, the cost to the customer is far below the cost that the utility has to pay to provide that electricity. Now, the tariff in Kenya is set up with this kind of three tiered progressive structure that has an advancing cost as you use more electricity. However, we're looking at median lines here and each of these median lines for all of these rural customers are below the 50 kilowatt threshold of the first tariff block. That means that for each unit of electricity, the customers are paying about nine and a half U.S. cents per kilowatt hour. In addition to this, they pay $1.50 per month fixed charge. So if we're looking at a customer with a 30 kilowatt hour a month fee, that means that they're paying this $1.50 a month fixed charge plus nine and a half cents times that 30 kilowatt hour. So roughly about $3.50, sorry, 30 times 10. Yeah, so roughly about $4.50 total for that entire month's electricity bill. Could we advance to the next slide, please? So you can actually advance all the way through and bring up all of the bullet points on this slide. Thank you. So what are the implications of this? So we're looking at grid customers here and we're seeing those typical bills that are around $3 to $5 per month. Now, that $3 to $5 per month is essentially what the utility can use to recover the cost of that connection. And utilities across the continent offer varying prices to customers based on policy for how much that connection actually costs. So in Kenya, for example, the cost is about $150 that's available to most customers for getting a new connection. However, on the other side of that is the cost that the utility bears for putting out the wires and poles and all the transportation costs involved in installing that connection. And so that cost across Kenya is averaging about $1,200 US. So the utility pays $1,200. The customer pays $150, which is often financed. And then that remaining over $1,000 in this case is borne by whom? It's often borne by the government. It's often borne by foreign multilateral donors and things like that. And so these are often coming in loans. But ultimately, this cost needs to get paid for somehow. And this represents some of the insolvency challenge that Frank introduced in the earlier slides. So one common strategy that's used to account for this discrepancy between the cost of the utility and the cost to a customer is really some cross-subsidization that looks at how we can use income coming from larger customers, often industrial customers or higher-consuming customers to offset these costs of investment into the grid. However, Kenya has about 3,500,000 industrial customers. And the grid needs to connect the next 5 million connections, so 5 million more connections. And there simply are not enough industrial customers. That would be on the order of 1,500 customers per industrial customer, 1,500 residential customers the cost of all of those connections per industrial customer. And that essentially means about a million and a half US dollars per industrial customer just recovered via their electricity bills. Now that's a lot of electricity bills. And not only that, that's the profit from those electricity bills. So minus the cost of actually providing that electricity. So we see some significant challenges here with the grid model. We can advance to the next slide. What does this look like for mini-grids? So Henry shared some excellent data from a set of mini-grids across Kenya. I actually have some data to share from mini-grids across Kenya and Tanzania that looks at some of the same questions. What is consumption like on these mini-grids? If we try this different model of using a distributed source and possibly keeping somewhat lower standards or somewhat different standards and reducing the cost of those connections, how much better can we do? If we can advance to the next slide. So this is data from 11 mini-grids across East Africa, across Kenya and Tanzania and looks at the monthly kilowatt hours per customer. So this is the same quantity that we were looking at for our previous slides where we saw that the median customers were in the roughly 20 to 30 kilowatt hours per month range. And so here we're seeing that most of these grids are actually far below that. We see a mean of around seven kilowatt hours per month and a median around two and a half kilowatt hours per month. So we're talking about one-tenth of that other medium we saw for rural customers. And most of these customers, if not all of these customers, are going to be rural as well. So we see that mini-grid customers consume much less than grid customers. And this has to do with the price of electricity. So the grids here are operated by an operator that's a private organization that is attempting to operate these grids at essentially either flat or a slight profit. And so the cost of electricity that they offer is far beyond usually in the $1.50 to $1.80 per kilowatt hour range. So quite high, but that's actually what they need to do in order to recover their costs efficiently. If we get advanced to the graph, the other challenge here is thinking about growth. So this is a graph that looks at the growth in electricity consumption from when these 11 grids started. And if we stay on that 1.0 line, the dashed line we see in the bottom of the graph, that is a grid that has not grown from the first month of consumption. But what we see as a pattern here is that most grids stay in the 1 to 2 range. They double over the 2.5 years that are represented in this graph at most. But many of them actually don't show very much growth at all. They're actually staying at less than 1.5. And some of those grids actually shrink their consumption after their first month. And so this challenge of how do we get these customers to not only improve their economic standing, but also improve their electricity consumption so that these grids can sustainably exist and sustainably grow in these communities is a severe challenge. And this is one of the big challenges for the industry and for microgrids in general. Next slide, please. So what does this mean and what can we do? So this growth and essentially it's flatlining that happens after a short number of years. It cannot sustain utilities and micro utilities. Whether you're the grid, whether you're a microgrid, you need customers to consume more of electricity. You need customers to buy more of your product to actually cover your costs. So the question is, as a grid planner, as somebody thinking about how we can provide electricity access to the largest number of people and end up with the best impacts from providing energy, how can we best set up this system? And we have a slew of tools at our disposal now, whereas it used to be just grids. Now we have mini grids. We have the ability to bring solar home systems and solar lighting systems that can bring to bear a lot of improvements on these systems. The question is how we can best deploy those resources. So one idea is actually doing better targeting of customers. So if we think about each of those lines we showed, they're actually all distribution. And if you're able to, in advance, best understand which customers are the most likely to consume more electricity, you can better target the type of electricity system that's available for those customers. So you could imagine if you knew that a customer was going to be a fairly low-consuming customer, then you could initially begin with a lower-cost source of electricity access, like a home solar system or even a solar lighting system. However, if you knew in advance that a customer was going to be a higher-consuming customer, then you could spend the large amount of money, the $1,200 US, necessary to get a grid connection to those customers. And so really doing better targeting of individuals rather than just targeting communities as a whole is a way to balance the heavy costs of providing these systems. And if you get that estimate wrong, it's simply adding that grid connection later. Now, one of the challenges, though, is how can we actually lower the cost of those mini-grids and grid connections? I believe our next speaker will talk quite a bit more about this topic and ways to really look at those costs. But this is a significant challenge. There's a lot of people looking at this. What I'll say is that one important factor here is that most of that cost is tied up in poles and wires. That's about 70% of that. So reducing the cost of those is challenging because they're not seeing the same sort of exponential decreases that we saw in solar that we are hoping to see in batteries as well. So as a researcher, what I can offer is what are some good further research topics in this area? And so one is really to understand how we can best emulate economic growth with electricity. Second, I want to discuss how prepaid meters affect consumption here. Poor reliability hinders consumption. Having a poor grid connection can be an enormous deterrent to people consuming electricity. Fourth is looking at technologies for reducing the cost of connections. And fifth is really understanding what the elasticity of demand is. As you change the price, how much power do people use? Next slide, please. Great. So thank you all very much. A short plug. I am always looking for strong PhD students who want to better understand the research methods that we can use for improving electricity access and reliability in sub-Saharan Africa. And I also want to acknowledge some colleagues, Simone Fobi at Vijay Modi at Columbia and Nathan Williams and Paulina Jaramillo at Carnegie Mellon University. Thank you. Thanks, Professor Taneja. Great job on kind of giving us an overview of the challenges of operations and specifically on the finance side of making these grids work over time. That's really interesting data. With that, we're going to pass it over to our final panelist, Omar Ghani, who is an entrepreneur with a passion for enabling energy independence, energy access, and eliminating the impact of climate change. He's the CEO of Kilowatt Labs. So Omar, take it away. Thank you, Frank. Good day, everyone. It's a pleasure to be here and to talk about the many challenges that we're facing in electrifying Africa. Thank you, Henry and Jay, for your very meaningful presentations. They address a lot of the issues that we face as we move forward. Our company is based in New York City, but our inventions were done in Dubai. We had a lab in Dubai. Our founder is based in Dubai currently, the inventor. He's from Pakistan as a MI. In fact, I'm calling in for Pakistan right now. I happen to be here on a business trip. And we have a lab in Dubai. We have a factory that's producing our product in China. And our first installations, in fact, have been in South Africa. South Africa, we have a very good partner there. We've established, we've installed some of our battery technologies there. Very shortly, we'll be installing some very good microgrids. So that's a bit about us. I just want to get into what we actually do. Next slide, please. Okay, so we believe that electricity access in Africa can be delivered by building microgrids that are based on 100% renewable generation, i.e. without dependence on fossil fuels and the grid. And we believe that the commercial challenges of doing this can be addressed with the right kind of technology. The business models depend on the output of technology, which is the output of electricity and the cost of generating electricity as Jay and Henry have explained. So when we started developing these products a few years ago, we did that with this in mind that we wanted to develop a system or a solution whereby we could generate independent power that could compete with fossil fuel energy, of course, and the grid in a lot of cases. And then, functionally and technically, our solution, which consists of two of our main products, which is the server and the storage, operates or delivers functionality just like the grid where you flick the switch a little bit and turn off the phone. You don't have to worry about load-level, you don't have to worry about it. Because at the end of the system, 4.25 per hour of power is the central resource system and everything. Next slide, please. Okay, so this is how the system works. We actually make two products. One is the Centauri energy server and I'll explain what that does in the subsequent slide. And we make a serious battery bank. This is a battery bank that's made from super capacitors. It's the first non-chemical battery bank that's made from super caps. So essentially, we connect the server and the storage together with PV panels. And because of the architecture of the server, the PV panels, they generate sunlight for four to six hours a day. The sizing of the PV panels is such that they are generating enough energy for 24 hours. So that energy is generated, fed into the server. The server then first directly feeds the load, which is quite different from basic microgrid design because normally PV panels feed the batteries and then the batteries go in and service the load because of the requirement of handling surge currents. But in the server, that's not required. So there's a lot of savings, a lot of optimization. So the panels generate electricity during the day. The first thing the server does is supply to the load directly. And then any excess energy charges the battery back. This happens for four to six hours a day, which is the seven hours or seven hours, whatever it is. And then when the panels stop generating, the server automatically switches to the batteries and starts supplying the load. And this circle is repeated every day. This grid on the left is actually a symbol for a redundant source. So the server can connect many sources. And if you have the grid, so the grid is connected, there is a redundant source and it's for days where the panels have not been able to generate electricity for 24 hours. So this is just a very simple overview of how the system works. Next slide, please. Another very simple slide, the Centauri and Cirrus-based microgrid can be installed in order to run communities, critical facilities, military batteries. It can be in a home or in a residential community. It can run factories, it can run offices. And of course, the rural communities, agriculture, mining sectors, islands, et cetera, et cetera. Basically, it can operate as a utility if required or it can be a home system as required. It has the flexibility to scale between the two extremes very easily. Next slide, please. Okay, so how does a microgrid look with a kilowatt-based system? First of all, these microgrids can run on 100% renewable generation. So you can have PV panels, the server, and the storage bag connected to the load, and that's all you need. Because of a lot of its functionalities, it is designed to operate every single device at the customer site. Unlike existing systems that are functionally restricted to operating critical items only. For example, motors and pumps are very difficult. A lot of oversizing is required if motors and pumps have to be run, but not with the server-based system. So it is a very simple connection to the customer's load, and it can run the entire load. Then it's a very simple plug-in system. So the customer or the installer does not have to buy any additional equipment. You just have to connect the PV panels to the server, the battery to the server, and you connect the output to the main distribution board of the customer and start the system. So the installation is easy. It's fast, it's expensive, it eliminates lines of failure, and it's inexpensive, rather, and it's very easy to do. So it can be done very quickly. The battery has a life of one million cycles, does not require any maintenance, and has the robustness to operate in very harsh environments and temperatures. It is because it's non-chemical, it is safe and not toxic. Because of all these reasons, the entire system operates in minimal supervision. This is a very critical requirement for electrification in Africa. It is a very critical requirement for electrification in Africa where maintenance and operations are very, very difficult to perform because of the distances involved. So it can be monitored remotely and controlled remotely, and it can be managed with very little supervision. Next slide, please. Okay, so let me get into what the actual products are. The energy server is a power electronics hardware and software platform that controls and operates the microgrid, and it delivers all the functionalities required to generate and distribute electricity from 100% renewable generation. So what are some of the key differentiating factors? First of all, because you, in order to generate 24-hour power in five to six hours, the PV sizing has to be four to five times. In normal systems, you actually need four to five times the inverters. You have to oversize the inverters, but with the server, you don't have to oversize it. So for example, a five kilowatt server, you can connect 20 to 30 kilowatts of panels. So this is a very important feature of the server. Secondly, it has the ability to handle torque load. It can handle up to 1,000% for two seconds, which enables the load to operate, to the customer to operate any load so they can start motors, pumps, compressors, refrigerators, anything air conditioners that have starting surge currents every time they start without actually oversizing the inverter, which is part of the server system. This is a very important quality in making the installation and running the entire system and the customer load very seamlessly and easily. Then it has plug-and-click connectivity of either a super-cap-based battery, which is what we produce, or a lead-acected family of batteries, or lithium-ion families. This further reduces costs because you don't have to buy a battery inverter or a battery management system. You just connect the battery bank to the server. The server has the appropriate module to manage the battery system. Then there's switching because different sources are being used. This is a very important function as well because it provides redundancy and it handles the intermittency of renewable generation delivering stable output. There's a busbar architecture which delivers switching between sources as far as the load is concerned in 0.0 milliseconds. Finally, it can take up to eight inputs, DC or AC, without any requirement for additional equipment, so you can connect PVE or wind. You can connect to the grid. You can connect a diesel generator or multiple diesel generators, and the system can be programmed to deliver sequential power or blended power. Finally, it has remote access capabilities, so you can monitor it, you can log it, you can control the operations remotely as long as there's internet. Next slide, please. The CITUS Energy Storage is the first super-cap-based energy storage that eliminates the challenges associated with chemical storage. Basically, we use super-caps as our storage media. We connect assembly of super-caps in series and parallel, and our technology is the control algorithms and the circuitry that controls the super-caps, and then our super-cap-based modules then deliver performance like chemical batteries. I saw a comment about using super-caps as batteries. That's exactly what we do. We have developed, we have overcome the limitations that super-caps have in actually delivering performance like a battery. I can get into that in the Q&A, but right now I restrict myself to the CITUS Energy Storage where it does, really. Because it's super-cap-based, it has very high DC efficiency, and it can be cycled 100% DOD, which allows for optimization of panel design because you're not losing energy in round-trip efficiency and you're not cycling fully. Then, again, another very important aspect, which is again based on the attributes of the super-cap, is that the rate of charge, the capacity and the life cycle of the super-caps or the modules, super-cap-based modules are not affected by the rate of charge, which allows, again, optimization of... The sun comes out and within one and a half hours the modules are charged because our solar-based storage modules are designed for 2C charge cycle, 2C rate of charge. Then because, again, super-caps have great temperature tolerance, so between minus 30 centigrade and 85 degree centigrade, you don't need auxiliary cooling or heating. Again, this saves a lot of money in terms of having to install auxiliary systems and also in terms of providing power to those auxiliary systems. Finally, all super-caps have 1 million cycle life, so our super-cap-based energy storage devices also deliver that kind of cycle life. Next slide, please. Okay, so what does this mean for microgrids? Well, what it means is that microgrids that are based on the Centauri and Cirrus systems, along with PV solar, can be effectively deployed across a variety of use cases in business models. The actual system regarding cost, so when you install a system that's PV solar and Centauri and Cirrus-based to give you 24-hour power just on PV solar, depending on the level of sunshine, if you're getting at least five hours a day for most of the year, it generates electricity anywhere between 20 to 25 cents per kilowatt hour for the life of the system, which is 25 years because we are limited for the life of the panels. And that is achieved by the fact that if everything is optimized, you don't have to oversize the inverters, you don't have to oversize the battery banks, you don't have to buy multiple systems and connect them together. So then you have optimal sizing, which reduces the cost. And then because the Cirrus is a supercap-based system, there is no chemical battery attached to it, it's just entirely on supercap. There is no replacement cost. Again, all this needs to put a relative fee because the technology eliminates oversizing, replacement, the total cost is a result in affordable electricity and then allows a lot of different kinds of business models to be implemented. And finally, it has allowed, because you're using 100%-based renewable generation, it eliminates the need for fossil fuel. Fossil fuel systems can be attached to them for redundancy for the, you know, out of 365 days if you have to run them for 10% of the time, it significantly reduces the greenhouse gases. Next slide, please. A simple overview. It can be used in all applications, the industrial, healthcare, agricultural education, residential, commercial, telecom, hoarding. In fact, we have a big project with MTN, which is South Africa's biggest telecom. They are using our storage modules to replace lead acid batteries. Next slide, please. This is a picture of the energy server and this is a 160 kilowatt machine. Next slide, please. And this is a picture of some of our storage banks. The one at the bottom is a 3.55 kilowatt hour, 48-volt module. The one on the right top is a 7.1 kilowatt hour, 48-volt module. And the one on the left is a 1.3 kilowatt hour module, which is now being deployed in various places. And that's the last slide. Thank you very much. All right. Thank you, Omar. Thank you to all of our panelists. I'm hearing some quiet applause in my head for the contributions of each of the panelists today. And with that, I'd like to get into Q&A. So a lot of folks are typing a bunch of questions into the Q&A for us. And some folks are even answering questions now into the chat. But the Q&A window, I'm just going to take some of these questions as we go through. A couple quick ones. Somebody asked, please define non-technical line losses. Maybe, Henry, you could take that one. I'm sorry. Can you repeat that, Frank? Just somebody's asking to define non-technical line losses. Oh, yeah, sure. Non-technical line losses are generally associated with theft, hence the non-technical aspect of it. It can be very large in many countries. India, for example, is one of the more notorious countries for non-technical line losses. Yes, absolutely. And that has a lot to do. Somebody else had asked a question. What are the main causes of the insolvency that I referred to in one of my slides at the very beginning? And according to the report by the World Bank, the four main causes are underpricing, basically pricing at a subsidized rate that doesn't cover the fuel cost or the cost recovery. Second would be transmission and distribution losses. That's both the technical and non-technical losses that Henry just spoke of, as well as bill collection losses in some countries. Even the energy that's not stolen that is billed, the bill collection rates are below 50%. And then overstaffing, which is basically hiring large numbers of people in regions where perhaps a more efficient operation could have fewer people. Another question from the audience here is who is developing the standards for AC and DC microgrids? I guess any of our panelists could take that one. Yeah, sure. I can take a stab at that. So there are a couple of organizations now that are working on standards for microgrids. In fact, IEEE and IEC actually have some that already exist. But I think there's been some criticisms as to whether or not these are exactly appropriate for the developing country context. So IEC is working on a low-voltage DC standard, as we speak, as is IEEE. And if you want to know more about those efforts, you can maybe contact me directly and I can get you hooked up with the groups working on those. Great. I also noticed somebody asked a question in the Q&A for Jay regarding his data. And Jay, you went ahead and put a link in the Q&A box. Could you explain what you did there? A question regarding what? I'm sorry, it cut out for a second. Okay. Yeah, it looks like you posted a link to our participants in the Q&A window. Was that in response to a question? That was. So that is data from the World Bank. Look, it was the slide that had the two maps of the continent, and it was data looking at rural and urban electrification. I will say just generally, one of the challenges here is data, getting accurate data, and also knowing how your data were collected. You can find these different sources, whether it's from the World Energy Outlook or World Bank or other sources for these data, but they often tend to be aggregated at country scale and can have tremendous amount of error in how they're collected if you actually follow the process. And so these data should really be a guide to begin your study, but if you really want to understand what's happening, it's important to be on the ground. It's important to work with the people that actually have the data that you're working with, whether it's in your organization or whether it's in another organization. But it's hard to get good narratives when you're working with aggregated data from other sources. Okay. And another question, I guess perhaps for Jay or maybe any of the panelists. The customers who haven't connected to the grid yet haven't developed consumption habits and may be more flexible in their behavior. Are you thinking of this in regards to demand response for microgrids? I thought this was a very insightful question. It's not just about, so my normal thinking is that microgrids and smaller constrained grids have more benefit from demand response, more benefit from altering demand based on what's happening on the supply side because they're constrained. Now, at the same time, your customers are actually more willing to deal with that flexibility as well. So I think that's a fantastic observation from the participant. And I think it's a very important point to realize. I actually have only seen a small number of microgrids that have tried demand response, but it's still been fairly limited. So it's something I'd love to see more understanding of how useful and how effective demand response can be for improving the financial sustainability of these kinds of systems. Yeah, let me tack on, go ahead. Let me tack on to that, if I may. Because a lot of these mini-grids are ultimately rely on batteries, any sort of demand response in terms of load shifting from the evening peak to during the day can save a lot of money in the, in how we design grids and the capital expense. So there's certainly an opportunity there. There's also an opportunity in increasing the diversity of the load, trying to keep it less coincident. Most data that I've seen that look at residential consumers of electricity, the peaks all happen at seven o'clock at night when the sun sets. Everyone turns on their lights and their TVs and there's not a lot of diversity. So you have to design your system around that peak and you don't get much diversity. So creative ways of spreading the peaks around also would make a big difference and there's a big opportunity there. Great point, yeah. And I will quickly add here that Segora International, the company I happen to work for, does have a proprietary smart meter that does enable the demand response and the type of flexibility that folks are calling for here. Going on to another question here for Omer. Where are the Centauri energy servers located? Are they central for each grid? Do you have multiple units distributed throughout the micro grid area? And then what is the approximate pricing? We're getting some questions on that as well if you are willing to share. Yeah, actually our product range starts at two and a half hours and goes up to a megawatt. So that's the information that we've done in the Middle East and in Pakistan over the last two, three years. This was before the server actually became the server it is. It's the version that we're using now. It varies from two and a half kilowatts to about 200 kilowatts and they are at the site level. So home, school, factory, commercial establishment, restaurant, you know, but all of the grid where these companies or individuals were using computer devices and didn't have electricity. So it can be sized based on your load. It can be installed either at the site or it can be installed for a community. If it's installed for a community, it would be connected to a distribution system so that it behaves like it is. And the pricing is about 40 to 50 cents a watt for the server. But remember, when you buy the server, you don't have to buy anything else. You just have to buy the PV panels and you have to buy the battery. Okay, great. So one last question, and I know people are really getting some great questions into the question window. I'm going to just, based on time, we said we'd keep you for 75 minutes and we're at 74. So last question is going to be somebody asked, could presenters discuss recent hurricanes and the wisdom of replacing grid-as-usual infrastructure, specifically in island context with these new microgrid models? A question about resiliency. Well, I would install microgrids obviously because they provide resiliency, they provide, especially in critical facilities, you need power. So these microgrids are very, very good for that. But the kind of devastation that's happened, it's very difficult to imagine PV panels or wind generators surviving. So it's always, it's difficult to imagine if there was a microgrid there that's based on PV. They can't sustain. No installation is designed for 200 mile-an-hour winds. And that's one of the challenges that we have to understand how to run. But installation do obviously need some input power. You don't have to have the generation source. And that's one of the things that we are always struggling with, especially in disasters. Jay or Henry, any last thoughts on resiliency? Yeah, I'll just say that, I mean, I agree with Omer and that it's really hard to strengthen against a hurricane. There's always going to be sort of a cost component that you need to be aware of. So if you want additional resiliency and reliability and all those things, then it usually comes at an added cost. So is it economically feasible? In some cases, yes, it's worth it. And I think for some other consumers, perhaps no. And Jay, any thoughts? Absolutely. I think that the challenge is not necessarily what happens during the hurricane. It's very difficult to deal with those situations. It's about resiliency. It's about recovery. And it's about how centralized that process is with PREPA in the utility in Puerto Rico. That process is entirely centralized. They've hired contractors from Montana that are a small organization that just doesn't have the manpower to build to install the same grid that we had before. Now, the challenge is how do you make a decision in order to fix up what was already there or take this opportunity to redesign and re-architect the whole thing and who gets to make that decision? That's a really, really critical challenge that puts us in a spot at this point in time. But building the same grid we built 50 to 100 years ago is seldom the right answer. So this could be a crisis as an opportunity situation, but at the same time you have millions of people in Puerto Rico suffering from a lack of electricity that there's no time for dithering. Yeah, absolutely, absolutely. And speaking from my personal experience, I spent most of last month in Haiti during hurricane season making sure that the grids that Segora operates in Haiti were prepared and ready to endure the hurricane. So, fortunately, it didn't take a direct hit, but there's been a lot of recent media coverage of the actions that we took to protect ourselves in that situation. And I have no doubt that microgrid infrastructure is exactly what's needed, in our case at least, for the resiliency and the ability to bounce back. So, with that, we're finished for today. Please, please go on, if you want to ask additional questions or if your questions weren't answered, you can tweet them to us at the hashtag is E4C webinar. You can also email webinars at engineeringforchange.org and our panelists will get back to you, as well as the E4C staff. As you can see on the last slide here, there's a PDH code for today's event. Go ahead and follow the link if you need PDH. And don't forget to become an E4C member and get info on upcoming webinars. So, thank you very much, everybody. Thank you. It's a pleasure to be part of the seminar. Yes. Thank you, everyone. Thank you, Alan. Please don't hesitate to email us to connect and find out more about these topics. We appreciate it.