The Metropolitan Council held a public hearing tonight on their draft Transportation Policy Plan. If you care about transit or transportation issues in the region, you should comment (you can do so through October 1). Here are four comments I have on the plan:
Our urban areas are significantly underserved by this plan. Even under the “increased revenue scenario”, we will spend $5 on transit to serve suburban commuters for every $1 we spend on transit improvements to places where transit makes economic sense (see here for my attempt at a geographic breakdown of projects). The Met Council, in the Thrive 2040 plan, has said they want to match transit service to the number of riders and intensity of land use. This plan does not do that.
The plan currently prioritizes projects like Gateway BRT (9,000 riders at $50,000 per rider) over projects like Hennepin Ave BRT (23,000 riders at $896 per rider). This is an example of how our urban areas (that are expected to grow significantly) are underrepresented in this plan.
It’s definitely not all bad. The Met Council for the first time has identified regional priorities for a bicycle network, which will give communities the ability to apply for funds to upgrade their local network if it matches the regional plan. Many of the transit projects identified are much needed improvements (Hennepin, Chicago, West Broadway), but are simply not adequately prioritized.
Over at streets.mn, I ask some questions about the Met Council’s new northeast metro water supply plan. Here is a big one:
Where is the conservation alternative? The cost and feasibility of reducing water use are not analyzed as part of the report. Building nothing and simply asking/incentivizing/requiring people to use less may be the cheapest option. According to the report, water use in 2010 was 92 gallons per person, per day in these communities. The ratio of peak day demand to average day demand ranges from 1.7:1 in Forest Lake to 5.9:1 in Lexington. The report hints that this is “mainly attributed to irrigation and outdoor water use needs”. Sprinkling lawns in other words. Many options exist for conserving (potable) water – from retrofitting toilets, sinks and showers, to using captured rainwater to irrigate, to simply paying people to remove lawns and replacing them with low-water alternatives. For the cost of the alternatives to serve all northeast communities with new water supply (~$600 million), you could pay every household over $1,400 to remove lawn, and keep paying them $40 every year after that. Without an analysis of conservation alternatives, this report seems inadequate.
Solar PV seems to be the current darling of the renewable energy world. But how much “resource” is really out there? How much should cities rely on the development of local solar resources to meet their climate and energy goals? What trade-offs should urban cities make between desirable things like tree canopy and maximizing solar energy resources? GIS tools and new data resources can help begin to answer that question.
Counties and states are beginning to produce LiDAR data more regularly, which provides the building block information needed to analyze solar resources on buildings and elsewhere (see my previous post for a brief intro to LiDAR, or see here). Minnesota happens to have LiDAR for the whole state, and Minneapolis has a climate action goal that references local renewable development, so I’ll focus there.
So how much solar electric potential does Minneapolis have? Enough to supply 773,000 megawatt-hours (MWHs) each year, at the upper bound. That would mean covering every piece of rooftop with good sun exposure and appropriate pitch (southeast to southwest facing or flat) with the best modern PV panels. It would also mean solar installations on 68,351 structures, consisting of over 2.3 million individual panels.
773,000 MWhs would represent about 18% of Minneapolis’ total annual electricity consumption (based on 2010 figures). It would be the equivalent of reducing 392,684 metric tons of CO2 (also based on 2010 figures), which is equal to the emissions from the energy usage of almost 36,000 average American homes each year.
There are some limitations to this calculation, and some additional interesting findings, but first a brief description of how I came up with these numbers.
I briefly covered how to calculate solar potential in a previous post, and the process for this analysis was similar. I was able to get my hands on the solar insolation raster for the whole city thanks to the excellent work of some students in Dr. Elizabeth Wilson’s capstone class at the University of Minnesota’s Humphrey School. Solar insolation represents a measure of the total energy from the sun reaching any particular point (each square meter in this case) on a building, tree, earth, etc. To calculate this, ArcGIS has a complex tool called Solar Radiation Analysis. It takes in to account things like how trees shade buildings, and how the sun moves across the sky at different times of year based on the latitude of a particular point on earth. It spits out a measure of solar energy hitting that location over the course of a year, measured in watt-hours per square meter. This gives you a good idea of where exactly on each building a suitable spot might be for a solar PV system.
LiDAR data can also be used to calculate the slope of roofs, another important piece of information to understand solar potential. This allows a user to pick out areas of flat or south-facing roofs.
Finally, Minneapolis supplies building footprints, so I knew approximately what was a roof. I confined my analysis to building roofs, assuming we don’t want any of our precious open space filled with solar panels. I also buffered the roof edges, since I’m told OSHA requires some open space between the panels and the roof edge for safety, at least for flat roofs. I also considered 1,000 watts to be the minimum size that would warrant an installer to climb onto a roof.
Combine all this with some assumptions about the space needed for installations on flat and sloped roofs (the students helped with that too) and information on the size and power output of panels, and you get a measurement of the total “good” roof area and associated potential energy production from each roof.
That’s enough how-to, here are more interesting findings.
The 100 buildings (0.14 percent of the total building with solar) with the largest solar potential would provide 14 percent of the total production, or over 109,000 MWhs annually. The 1,000 buildings (1.4 percent of the total buildings with solar) with the largest solar potential would provide 43 percent of the total production, or over 333,000 MWhs annually. Targeting these structures for further analysis and possibly incentives would probably make sense to achieve the largest economies of scale for installation costs.
The 100 highest-potential buildings are geographically concentrated in roughly three areas: the northeast industrial area – roughly north and east of the U of M campus, the Lake Street/Greenway Corridor, and extending from the North Loop along the river into north and northeast Minneapolis. Unsurprisingly, these are areas that still have many large, flat-roofed warehouse and industrial buildings. If Minneapolis wants to maximize its solar resource, we may want to think about the trade-offs in redeveloping these areas or developing high density near them that may shade existing rooftops.
Commercial, industrial and single-family residential structures (based on parcel data) each account for almost exactly 23% of the total roof-top solar potential in the city. The next largest potential was among apartment properties at 9%, and duplexes at 7%. While the top three were evenly split potential-wise, single-family residences with good solar potential included over 46,000 structures, while commercial and industrial together was about 4,300. See economies of scale note above.
The fact that 46,000 residential structures have good solar potential means that lots of homeowners, even in leafy Minneapolis, could be empowered to go solar. This would be a more powerful political constituency than a small number of commercial property owners. Obviously some would face the trade-off between more trees and their benefits and electricity from solar.
Suburban areas are much more likely to approach energy production equal to energy usage. With its high density commercial core, Minneapolis uses a lot more energy than it can produce on its roofs. Residential structures are also smaller and more shaded than many suburban areas. This isn’t necessarily a bad thing, as density brings many other environmental benefits, like the ability to use transit cost-effectively.
Xcel Energy limits the size of solar installations they allow to be connected to their system. An interconnected solar PV system cannot be designed to produce more than 120 percent of the customer’s total usage from the previous year. Many homes in Minneapolis, and possibly low-energy warehouse buildings, could accommodate systems larger than that. This analysis limited system size only based on roof/sun conditions, and not electricity usage in the structure since that wasn’t known. In some cases, this means this analysis over-represents solar potential.
This analysis includes no information on roof age or structural integrity. Some flat-roofed buildings aren’t structurally able to accommodate solar without expensive retrofits. Residential structures may need to have old roofs replaced before putting on a solar energy system (which are typically designed to last 20 years). Some structures, like parking ramps and stadiums, would require additional structural supports to be added before a solar energy system could be added. These factors could all further limit solar potential on Minneapolis buildings.
There was a geometry problem I couldn’t solve in GIS. While I could calculate the size of a roof area that got good sun and had the correct slope, I couldn’t quickly figure out how many solar panels of a certain shape (defined length and width) fit in that area. I only used total square footage divided by the square footage of a standard solar panel. Internet forums are filled with many people better at GIS than I discussing this problem (but not providing me with easy solutions). If anyone reading this wants to take a crack at it, let me know in the comments.
Today I noticed that my solar charge controller has been running for 100 days (it logs this among many other data points). Here are some highlights from the first 100 days:
The system has produced 32 kWhs from two 100-watt panels. This is roughly 2% of the total electricity consumption we saw over the same period last year.
Converting from DC current to AC current at low wattages is wildly inefficient. I usually run the wifi router and cable modem continuously off the battery and I lose about 40% of my produced energy to the inverter. It is much happier running closer to its peak (1000 watts). We should probably convert to DC.
Something happened to my charge controller settings when I converted to 24 volts. Although the controller was still charging, I lost about 10 days worth of data (hence the gap in the chart) and wasn’t able to communicate with it over that time. A firmware reboot fixed this.
Although very cold, clear days are when the panels perform their best, the sun just doesn’t shine for that long each day in January and February in Minnesota. The panels being on the ground doesn’t help either. Just from the middle of March to the middle of April I’ve about doubled my daily output.
All that said, this chart doesn’t really show total potential of the panels on a given day. If I didn’t use much of the battery the day before, panel production the next day was curtailed by the controller to avoid overcharging the battery. I’m trying to match the loads I put on the battery with the “capacity” of the season, but that’s sometimes tricky.
I recently learned we were accepted into the Minnesota solar rebate program for 2014! So with the help of a friendly solar installer, we should have a 2.8 kW grid-tied system installed sometime this year. Along with the grid-tied panels, the installer will be adding two panels on the roof dedicated to battery charging. Now I just have to wait…
I’ll start by saying I have strong feelings about Southwest LRT. So do some people on this very blog. You probably do too. However, I won’t be contributing further to the gallons of spilled real and virtual ink or weeks of public testimony. I’d like to talk about how we can set the stage for some other projects that could be really beneficial for transit-dependent and transit-interested communities. Nothing in this post should be interpreted as diminishing the importance of that LRT project, the upcoming decisions that will determine it’s fate/depth of its tunnel, or the correctness of any particular opinion about it. But I have this urge to start some positive conversations about other projects that need some support. Weird, right?
Many of the images in recent posts include “https” in the link, and since I’ve recently stopped using SSL, they seem to no longer work. You can still access the images by deleting the “s” from the URL. I hope to have a better fix soon.
Boston, New York City, Denver, Cambridge and other cities have created solar potential maps to help their residents understand that solar photovoltaic systems are viable in dense urban areas, and to demonstrate the potential that exists on rooftops.
Of course, I had to try this myself.
Minnesota produces LiDAR data, which is basically micro-scale elevation data produced by flying a plane back and forth in a grid and shooting the ground with lasers a bajillion times. Skilled/obsessive GIS users can clean from this data information that can be used to make a fairly accurate model of everything on the ground (buildings, trees, etc). GIS software also makes it easy to produce daily, monthly or annual solar insolation maps. By taking the position of the buildings and trees, knowing the latitude, and projecting how the sun moves across the sky throughout the year, the software calculates a total amount of solar radiation that will hit a point after shading, angle and other factors are taken into account.
After much tinkering, the Kingfield Solar Energy Potential map was born. The extreme density of the LiDAR data limits how large an area I could process (there were 4.9 million individual data points in this one small section of Minneapolis), but you get the idea. This map shows the area of each roof that might be appropriate for solar, how many panels could fit in that area, and an estimate of the annual production from those panels.
Some roofs are wholly inappropriate for solar, whether due to tree or building shading, orientation or size. But there is significant potential. If solar was installed on every appropriate piece of roof in this one-quarter square mile area, it would produce an estimated 2.2 megawatt hours of electricity each year, and avoid 2.9 million pounds of carbon dioxide emissions.
This little device is an ethernet to wi-fi adapter. It connects my solar charge controller to my home wi-fi network so I can make fancy graphs. It uses 1.2 watts per hour. I know this because I measured its usage using a watt meter. I do this with everything I power from the solar batteries.
I have a hunch that this is what solar does to you, makes you compulsive about energy use. Even if (when?) I have a large grid-tied system, I imagine myself checking the daily output, and constantly thinking about how to reduce my usage to match.
On very cloudy days, this little thing has used over 45% of the energy produced by the panels. I unplugged it. For now, graphs only on special occasions.
As part of my plan for the eventual expansion of my off-grid solar energy system, I recently added a new charge controller with Maximum Power Point Tracking (MPPT). Besides being much more efficient, this controller is capable of producing reams and reams of wondrous data, and is network-connected, meaning I can geek out on battery voltage and array current from anywhere in the house! The charge controller I had was great, but it wouldn’t handle anything beyond a few more small panels. Now I should be able to go all the way up to 750 watts of panels (my goal). So, thanks Santa!
While installing the controller, I also took the opportunity to install a breaker box, which should bring me closer to code, and upgrade to larger diameter battery cable, which should reduce efficiency losses.
The MPPT advantage
MPPT is a fancy way of saying the charge controller is able to send significantly more energy to the batteries from the same panels. How much more? After only a few days of testing, I estimate 40 – 60% more than the Pulse-Width Modulation (PWM) controller on days when the battery is low. (If you want to know the details of how MPPT works, I found this explanation helpful.)
Here’s some actual data from my system which I think illustrates the MPPT advantage well:
The blue line is the amps, or current, coming from the panels. The red line shows the amps the controller is putting in to the battery. It’s higher! The magical MPPT doohicky converts excess voltage into amperage (remember, amps X volts = watts) so less of your panel’s potential is wasted. On this particular day, I estimate the charge controller may have been able to wring an extra 100 – 150 watt-hours from the panels.
There are other interesting things going on here, so here’s a little annotation:
Here’s the next day, when the battery starts out the day almost totally full. It was very sunny.
The controller limits the array current and current to the battery significantly because the battery is almost fulled charged. The gentle downward slope in the amperage is a function of battery charging called absorption. Less current is pushed into the battery as it reaches capacity.
I can track hundreds of days of watt-hour production, so I’ll do another update when I can show some seasonal changes. How I yearn for the days when the panels get more than 4 hours of sun per day!