Showing posts with label rooftop solar. Show all posts
Showing posts with label rooftop solar. Show all posts

Sunday, August 6, 2023

A Weird Period With the Solar System Appears to End Well

I've been quiet about the status of my solar system for a few months now, and there's a good reason for that. Sometime in May, things got weird, and it took until late July to un-weird them, and the result of that process didn't become apparent until, well, today, in early August. 

We entered May feeling optimistic. We'd seen our first power bill charging only for delivery and fees, not grid consumption. At the end of a period with sunny and dry weather, we expected more sun and more days producing more power than consuming, or at least offsetting high consumption from the use of AC and the pool heater. Yet in that same peppy post from May 1, I noticed that production in March was higher than that of April. In hindsight, I should have seen that as a red flag. 

Things got weird in the month of May. Production came in below April and by a lot: 731 kWh versus 865 kWh. That by itself is weird. Days get longer in May, and the weather generally gets sunnier as summer approaches. It really should have been accelerating, not slowing down. I reckoned the weather was a factor: It was an unusually cloudy month for May, with several rainy days in the second half of the month.

I also suspected that perhaps the panels were dirty. The same tree pollen that coated my car with a fine dusty patina appeared also to coat the panels. (You can kind of see it in the picture of the back array at left.) In one funny moment, I took my water compressor and tried to manufacture some "rain" on the front panels. It was fun to try, but it made no difference.

As the month progressed, my concern mounted. I put in a call to our vendor, Kasselman Solar, who sent out a service rep to take a closer look. (She quickly dismissed my theory about the pollen dust.) Communication, it seems, between one of the basement inverters and its home base stuck out right away as an apparent problem. The location of my home has exceptionally bad cellular phone coverage, and the SolarEdge inverter relies on a cellular modem to report its production and output, including its exported power to the utility grid. That old saying "proof or it didn't happen" applies here. If the inverter can't report exported power, I don't get credit for it on my bill, end of story. And those credits are essentially the reason the entire system exists. 

One of the two inverters, it turned out, wasn't communicating at all. It had a faulty comms board, the internal part that talks to the cellular modem. Either the board or the inverter itself would have to be replaced. 

By mid-June, a new inverter had been swapped in and all seemed to be on its way to resolution. Yet as June swung into July, production still didn't rise to a reasonable level. Weather continued to present a challenge. June and the first week or so of July were marked by numerous severe thunderstorms and flooding in the Hudson Valley. I couldn't really expect high production numbers under those conditions. (That dust I had worried about? Totally gone.)

Yet as the weather conditions improved, the needle still didn't move meaningfully. In July of 2022, the first full month of the system's production, the system produced 1.43 mWh. In July of 2023, I produced 732 kWh, or about half the level from the year-ago period. Something was clearly amiss. 

A second service visit from Kasselman revealed the problem: In the technician's argot "One of your strings is down." It meant — I think — that a line connecting some of the panels to the inverter was not working. The source of the problem, it was later revealed, was a "bad crimp," which again I think meant either a faulty component or a poor connection. "Bad crimps happen," he said. Okay then. He and a colleague climbed up on the roof, fixed what had to be fixed, and that was that. Problem solved? Not just yet.

The communication problem was still unsolved and appears as of today to have been more significant than I realized. In parallel to all this I had been struggling with steadily worsening Wi-Fi around the house and had decided to upgrade it. Wherever I live, I make a point to get the fastest Internet connection I have, and so I have a 1-gigabit plan from the local cable company.

But the trio of TP-Link Deco M4 Wi-Fi routers the cable installer had suggested just weren't getting the job done. They seemed overwhelmed by all the connected devices in the house: Three phones, three iPads, a smart TV, smart lights, a connected washer, dryer, dishwasher, oven, and fridge, three always-on computers, network-attached storage, a smart speaker system. You get the idea. 

I went looking for a more robust mesh router system. I turned to Wirecutter, the product review team at The New York Times, and read their reviews of wireless mesh systems and settled on the Asus Zen WiFi AX. I bought a set of two and liked the results. (My cable connection is still wonky, but I'm hoping for an upgrade to fiber optics sometime soonish.)

In anticipation of the second visit from Kasselman's technicians, I bought another set of two. All in the upgrade cost me $700. It seems to have been worth it.

In a conversation with the Kasselman service manager, I learned that the SolarEdge inverters are not approved by the FCC to connect to Wi-Fi, but if a router is close by, a technician can connect it to the Internet via an ethernet cable. This is where the Asus routers made a difference. The main router creates a 5 GHz "backhaul" connection to feed the other Wi-Fi nodes stationed around the house. One of the new routers, I reasoned, if placed in the basement and connected by ethernet, would give that inverter a clear no-excuses connection to report its status. On that second service visit, after the fix on the crimp and the string, I had that fourth router powered up and ready for them to connect to the inverter. Now it's sitting on top of the home battery. Eventually, I'll mount a shelf there. 

Today is the first day that it became clear that we're in a different phase with the solar system. The data connection is working, and SolarEdge has updated its monitoring apps, and the data is just looking different than before. And by different, I mean, better. So much better. 

Last year, the most productive days of the summer peaked at about 60 kWh. This week we've already seen two days with production north of 80 kWh. As I write the sun is setting on the third day this week with production north of 82 kWh, of which more than 40 percent has been exported to the grid. And since temps have been in the mid-70s, our use of the AC has declined, keeping overall consumption down. 

It almost seems like the entire system has been running at less than its full capacity since installation. I'm going to try not to think about all the missed sunlight over the last year, and focus instead on the prospect of improved performance ahead. If this is the new normal for sunny mid-summer days then that is very good news indeed. 

One thing I definitely learned: If you live in an area where cellular coverage is anything less than ideal, do whatever it takes to get a hardwired Internet connection to the inverter. That data link is a lot more important than I ever realized. 

Tuesday, November 1, 2022

The Power Grid Failed for a Few Hours on Sunday. Here's What Happened.


On Sunday a little after 2 PM I had taken my daughter out for a lunch of grilled cheese sandwiches when my wife texted: "The power just went out."

The power has "blinked" a few times during the night in recent weeks. We notice when certain appliances in the bedroom light up after coming back on. It's annoying, but not so frequent as to rise to a serious problem. I thought this might be another instance of that. It wasn't.

At 2:21 PM our local utility Central Hudson Gas & Electric Corp. (CenHUD) texted to inform us that "We believe there is an outage" affecting our address. The message estimated that 608 customers were without power at that moment and that it was expected to be restored by 5:30 PM. 

My daughter and I finished our lunch and drove home. My outdoor lights — both at the front porch and outside the garage doors — were lit. And inside, my wife was watching TV as if nothing were amiss. To her nothing was. Yet there was no indication from CenHUD that the power outage had been fixed.

I opened up the SolarEdge monitoring app on my phone and saw the information displayed at the left. It showed that my solar panels were generating 330 watts of power, all of it flowing directly to my house. The TV was working normally as were the cable modem and Wi-Fi routers that supplied it with content. Lights came on normally in the bathrooms and the kitchen. Had I not received the text message from CenHUD, I might have been unaware there was a problem with the power grid at all. 

As far as I could tell, all the power we were using at that moment was coming directly from the roof panels themselves, and nothing was coming from the battery. I explained all this to my wife, who was suddenly rather impressed. In the months before we signed the papers to install the panel-and-battery system I had made myself something of a pest on the subject of solar power. Now it was paying off. As the implications of the moment dawned on her, she slapped me a high five. 

I settled into my office to watch an MLS Playoff game on my iPad. My daughter watched a movie in the family room. Life proceeded as normal. 

While I was watching the game, I kept an eye on the SolarEdge app an outage map that CenHUD had provided. 

The estimated number of affected customers grew to about 1,000 and the estimated time to resolution went from 5:30 PM to 6:30 PM. No explanation for a cause was given, but it was likely routine. A faulty line or a blown transformer. 

But I started to wonder what might happen if the outage dragged on into the night. Saturday was a sunny day that started cold with temps below freezing but ended in the mid-60s. The sun was due to set just before 6 PM. If the outage were not resolved, we'd be running on battery power, something we've yet to experience. 

Fortunately, we didn't have to find out what it's like on battery power. CenHUD corrected its outage on schedule and our connection resumed before sunset. 

I went back into the monitoring app the next day and looked a little more closely at the data. 

It's clear from the data that prior to the grid failure, the panels had been producing substantially more power than the house was consuming at the time. And in fact, had we upped our consumption, say to bake a cake, or had we wanted to turn on the air conditioning, we would have been fine to do so. 

Since it was a sunny and clear day, at the moment of the grid failure, the panels had been generating 6.2 kWh worth of power, and through the inverter, exporting most of that externally to CenHUD. Our consumption at that moment was less than 1.2 kWh. What I think happened is that when the inverter temporarily lost its connection to the grid, it simply stopped sending out that excess power. 

Obviously, we were lucky that this outage happened on a day with sunny conditions. Power grids fail during severe weather events all the time. Part of my calculus in adding the solar panels and the battery has been to hedge against these moments and improve our chances of easily lasting through these moments with an easy source of power plus battery backup. And I don't expect everything to work as it did on Sunday. I've been told that when running on all-battery power during the night, we'll want to be careful with our consumption, and avoid high-consuming activities. But as yet I've no experience against which to gauge that advice. 

Meanwhile, I have an update on October's performance. For the entire month, we produced 811.87 kWh against total consumption of 515.48, leaving us with a production ratio of 1.58, which is both higher than I reported here on Oct. 19 and also higher than the level of 1.25 I recorded for the month of September. I had hoped to finish with that ratio above 1.6, and before some rainy days last week, it had indeed risen above that level, only to settle lower.

I exported 570 kWh or 70 percent of the total amount produced, and that added up to more than twice the 274 kWh I imported from CenHUD's grid. Overall I produced nearly half or 47 percent of the power I consumed and imported 53 percent. And despite Sunday's brief outage, usage of the battery remained at zero. 

I still haven't seen a bill from CenHUD against which to compare these results, though one is expected any day now. That will be the subject of my next post. 

Wednesday, October 19, 2022

It's been four months since my solar system was installed. Here's how it's going.

The system went live during the mid-afternoon of June 17, which means that as of this writing, I now have about four months of operational data upon which to draw some early conclusions.

First, let's start with the big numbers. In four months, the system has generated 4.81-megawatt hours of power. During the same period, my house consumed 6.42-megawatt hours. Conclusion: In four months I generated 75 percent of the power I consumed. So far so good. 

But there's more to understand. During the summer months, I was running the air conditioning, and on some days the pool heater. AC use was fairly constant throughout the summer. The all-electric pool heater, I quickly learned from watching the data in the tracking app provided by SolarEdge, the manufacturer of the basement inverter, raised the overall electrical demand significantly when it was in use. Below is the graph for July's consumption (red) and total production from the solar panels (green) and self-consumption (blue). 

As you can see, the pool heater was running full blast all day on July 1 and not as much during the majority of the month. July was hot, and once the water in the pool reached its desired temperature, it didn't require much effort from the heater to keep it there. I also made a determined effort to keep the heater turned off during the month because I was a little surprised at how much power it consumes. All in, during July, I produced a little more than 57 percent of the power my house consumed. In August, that figure rose to nearly 63 percent. 

And then came September. It was cooler than expected and despite several days with temperatures in the mid- to high-80s, the overall need for air conditioning declined, as did the desire for comfortable water in the swimming pool. We had it closed for the season on Sept. 15 and rarely turned on the AC to cool the house. Data from SolarEdge shows that my system produced 1.02 megawatt hours against consumption of 821 kilowatts. In short, I produced 1.25 times the power I consumed. See the graphic below. 


This aligns with an early hope I had for the system before it was installed: That New York's sunny but cooler days of late summer and early fall would work to our advantage. I was right. October has been even better: As of today, the production-to-consumption ratio has risen to 1.55 with more than two weeks to go in the month. 

I'm curious about production during the winter months. For one thing, the days are shorter. The sun hits the panels progressively later in the morning and leaves them earlier in the evening. It's worth comparing a productive day from June to one in October.


This graph is from June 25. The panels first begin to produce at about 5:45 AM, and did so consistently until about 8:15 PM. The pool heater was off, but the AC was cycling on and off, as evident from the red spikes in consumption. This was the most productive day in June, incidentally, and it ended just short of breaking even against overall consumption. For the record, the official sunrise occurred at 5:22 AM while the sunset was at 8:33 PM. 

Now take Oct. 6, which is pretty close to the most productive day I've experienced so far this month. 



As you can see, the system produced more than twice what the house consumed, but its window of production was more than four hours shorter, starting at about 7:15 AM (official sunrise 6:57 AM) and ending shortly after 6 PM (official sunset, 6:29 PM). Obviously what I'll be watching is how the shorter days of winter narrow that production window even further. And then there's snow to consider: Panels that are covered in snow can't produce electricity. 

There are also a few other things I have yet to fully understand: How does exported power affect my utility bill? The green spaces in the graphs indicate when the system is producing more power than the house is consuming. That power is then "exported to the grid," and the numbers on my utility meter run backward. I've been photographing the meter readings several times a week to track the progress and can confirm the numbers are somewhat lower now than they were from their late August peak. But I don't yet fully understand how this affects my bill, in part because I'm now billed on a bi-monthly and not monthly basis. I'll write more about this as the puts and takes clear up. But overall I'm liking what I see.