One constant in my life is every morning I brew up a pot of coffee. One day I decided to brew it with solar power. The result was an adventure in minimal system design. This article is meant to be read with a bit of humour.
Design
My load analysis amounted to (7.5 amps X 120 V) approximately 900 watts of AC power needed for 15 minutes a day. (coffee is transferred to a thermos after brewing). Load analysis done, this meant I needed a 1000 watt inverter. A friend had a used Xantrex DR1524 available so I bought that. This meant that I now had an inverter that could put out 1500 watts and that would draw power from a 24 volt battery bank.
Battery bank analysis. 900 watts X 0.25 hours/day = 225 watthours/day power draw. Inverter Efficiency vs Output Power charts show the inverter will have over 90% efficiency at 900 watt output. Therefore to get 225 watthours out of the inverter daily I'd need to draw 250 watthours (250 in X 90% = 225 out) from the batteries. 250 watthours/day from a 24 volt battery bank calculates out to be 10.4 amphours/day (250/24). To get 900 watts output from the inverter requires 1000 watts input (1000 in X 90% = 900 out). 1000 watts out from a 24 volt battery bank calculates to be a current draw of 42 amps (1000/24).
Maximum recommended battery discharge rate is usually quoted as a C/5 rate, meaning a current draw (amps) of no greater than 1/5 of the battery bank capacity (in amphours). The Ah capacity needed for my system would then be 210 Ah (42 X 5) to not exceed a C/5 discharge rate. Rounding up rather than down, I therefore would need 300 Ah of battery capacity.
My friend had some ReadyWatt RW-100 AGM no-maintenance batteries for sale. These batteries are 12 volt and rated at 100 Ah each. The Xantrex DR1524 inverter is designed to draw power from a 24 volt battery bank. For 24 volts I would need two batteries in series. For 300 Ah capacity I would need 3 parallel strings at 100 Ah each. 2 in series X 3 in parallel = 6 batteries needed.
Six batteries needed for a load of only 10.4 Ah/day would seem to be excessive. At 100 Ah capacity each, the batteries would seem to be good for 100 Ah X 50% drawdown = 50 Ah/day, far more than the required 10.4 Ah/day. This is an interesting first lesson in minimal load design: a big 15 minute load translated into formulas designed for hourly load calculations results in oddities like 42 amp current draws but only 10.4 Ah used per day. If I had not caught the current draw calculation and instead had built the system with just 100 Ah capacity to handle my 10.4 Ah per day load I would have been drawing down the batteries at a C/2.4 rate (100/42) which would have worn them out quick.
The ReadyWatt RW-100 batteries are AGM no-maintenance batteries which is just what I needed as then I do not have to worry about maintenance while travelling. Also, since my system is installed inside my house, there would be basically no toxic/explosive fumes venting off from the batteries. AGM batteries have a reportedly low self-discharge rate, and true to such reports, my batteries still held 12.1 volts each after sitting idle for two years waiting for me to put this all together. (12.1 volts is about 50% discharged, but not bad for two years sitting around doing nothing).
My battery bank has a capacity of 300 Ah. With a 10.4 Ah/day load, my battery bank undergoes a Depth of Discharge DOD of only 3%. (10.4/300). At a targeted maximum DOD of 50% I would have of 14.4 days (300 X 50% / 10.4) of no sun days autonomy and still be able to make my coffee. At full charge the 42 amps output from the batteries represents a discharge rate of C/7.1 (300/42). The targeted maximum discharge rate of C/5 would occur when my batteries had 210 Ah capacity left. My battery bank therefore gives me a discharge rate limit autonomy of 8.6 days (10.4 Ah/day X 8.6 days = 90 Ah capacity used up without recharging), beyond which my 42 amp current draw would exceed a C/5 discharge rate from the Ah capacity remaining in the battery bank. After 8 days of making coffee using 10.4 Ah/day with no sun to recharge the batteries, I would have 217 Ah capacity left in the batteries, and my total DOD would be 72% (217/300). This is the second interesting lesson in minimal load design: the difference between Depth of Discharge days autonomy and C/5 discharge rate limit days autonomy calculations.
Another battery bank consideration when designing a minimal load system is that most batteries are rated at a C/20 discharge rate (p.63 SEI Photovoltaic Design and Installation Manual). Gel cell batteries (AGM is a type of gel cell) should never be charged at a rate higher than C/20 (ibid p.64). With full sun on the panel, this system would deliver 7.5 amps charging current (roughly 200 watts X 90% charger efficiency / 24 V batterry bank voltage), which is a charging rate of C/40 (300/7.5). If a system takes power out of a battery bank at a rate greater than C/20, such as the C/5 this system design uses, the actual Ah capacity of the battery bank is less than the rated capacity (ibid p.64). In this case the reduction in actual battery bank Ah capacity to rated battery bank Ah capacity may be around 20%. A 20% reduction in rated Ah capacity would reduce the actual Ah capacity of the battery bank in this system to 240 Ah as compared to the 300 Ah rated capacity. Very technical considerations such as this is why, when designing a system, one rounds up instead of down, as above, when I calculated I would need 210 Ah battery capacity and then rounded that up to 300 Ah capacity for the system instead of cheating the system design down 10 Ah to make for a 200 Ah battery bank and therefore save money on batteries. The daily load of 10.4 Ah of this system is very low and so operation of the system would not be affected by this technical dip in actual vs rated battery bank Ah capacity when the coffee maker is on. The DOD days autonomy calculation would be reduced to 8.7 days (300 X 50% = 150, 240 -150 =90, 90 / 10.4 = 8.7) and the C/5 discharge rate limit days automony calculation would dip to only 2 days (240 - 210 = 30, 30 / 10.4 = 2.9). This C/5 design versus C/20 rating consideration factor seems to be very important. Plus then one could also consider the AGM factor that AGM batteries, because of how they are built with the glass matt between plates, which is a hinderance to electron flow between plates, are even more sensitive to this C/5 design versus C/20 rating 20% derating factor as be for lead-acid batteries. In this system where the coffee maker pulls a 42 amp current draw out of the batteries, a C/20 discharge rate of the battery bank would call for a battery bank with 840 Ah capacity (42X20), resulting in 9 parallel strings of two batteries each for a total of 18 batteries. 18 batteries for a 10.4 Ah daily load is clearly impractical. Such is another interesting lesson in minimal system design: all the various individual design considerations potentially balloon in combined importance as compared to a bigger and more general whole house design. This also hilites where the experience factor of the system designer comes into play, knowing what works in a practical system from past experience with different types and brands of batteries as combined with different loads. Will stick with the 6 batteries for now and see what happens.
The next consideration in design was power from the PV panels. 250 watthours/day gets drawn from the battery bank (24 X 10.4). Battery bank efficiency is usually calculated at 90%, meaning power output is 90% of power put in. Also the system would need a charge controler, again with a rough 90% efficiency. Therefore to get 250 watthours/day out of the battery bank I need to put in 308 watthours of power (308 X 0.9 X 0.9 = 250) from the PV panels on a daily basis.
In central Alberta one gets 4.3 average Peak Sun Hours per day annually and 1.6 average Peak Sun Hours per day in the worst month in winter. As I want to brew coffee in the winter too, my system design must focus on the worst month. With 1.6 average sunhours/day in winter and 308 watthours/day needed from the panels, I would need 192.5 watts worth of panel power (308/1.6=192.5).
I bought a single Sanyo HIP-200BA3 panel. This panel is rated to put out 200 W at the Standard Test Conditions STC of 25C. In winter, with the worst sunhours, the temperature is also much lower than 25C. The panel literature tells me that I gain 0.172 V output per degree C less than 25C. Therefore at winter temperatures the panel will put out more power than the rated 200 watts. This is great and will help somewhat in offsetting the lesser total sunhours available. The panel is rated at an output of 68.7 Voc at STC. At -40C this can be 79.75 Voc. (68.7 V + 0.172 V/C X 65 C [difference between STC at 25C and -40C = 65C]).
I needed a charge controller that could take an 80 V input (Voc potential at -40 C) from the panels and step it down to 24 V output to the batteries. The Outback MX60 was recommended. The voltage windows were right. The MX60 can handle up to 141VDC in and steps down to various voltages. Plus it has MPPT tracking features. Sold.
So there it is, the system design. A single Sanyo HIP-200BA3 panel feeding into an Outback MX60 charge controller which feeds into six (2 in series and 3 strings in parallel) ReadyWatt RW-100 AGM no-maintenance batteries which then feed up into a Xantrex DR1524 modified sine wave inverter which supplies the 120VAC needed. In the worst winter month I will get, daily, on average, pretty close to exactly the power needed to make my coffee. I will have 8 days of autonomy in regards to a design maximum C/5 discharge rate for the coffee makers power draw, which would result in a 72% DOD after 8 days.
Installation
Skipping what all went alright, I'll outline the trouble bits. First, the Sanyo panel has a unique mounting lip, as mentioned in the product brochures, as compared to most panels with square frames. The clamps I later bought did not fit right. To make all fit I cut the lip on the panel frame. Not a problem, but doing so perhaps voids the warranty on the panel.
Second, finding cable lugs for the battery cables. #2/0 AWG cable lugs with a 3/8 hole in it, as be the terminals on the batteries, were not in the supply catalogs of local electrical shops around here. Eventually I used a combination of cable lugs and some big aluminum cable clamps for others.
Third, finding the right DC disconnects and fuses needed. What I ended up doing is probably not exactly up to code. Between the panel and charge controller I used a 15 amp fusible disconnect and built my own box for it. Between the charge controller and batteries I used a pull out disconnect, no fuse [fuses protect wires, PV panels are current limited, hence the current from charge controller to the batteries is limited and would never exceed the capacity of the #6 wire used].
For the very important DC disconnect between batteries and inverter I cheated and bought a MidNite Solar E-Panel which contained a 250 amp DC breaker. The 250 amp breaker fits code requirements for #2/0 AWG cable in "Free Air", but brings up point four, that the MidNite Solar E-Panel has two parts to it, a top part with breakers for wiring in a generator or 120VAC home power to the inverter for battery charging (a feature supported by the inverter), and a lower part for the DC breaker and wiring. Turns out that in Canada the electrical code does not allow for AC and DC to be together in the same box. Inspectors here require one to construct a metal separation plate between the two powers in this box. The AC wiring bit is not connected in yet in my system and probably never will be, as this project is an experiment about power from the sun and there are no critical loads. As I really only needed the breaker and a box for it I probably could have saved some coin here on this bit.
Start-Up
Start-up went smooth. With AGM batteries it turns out that almost all of what is written in the MX60 instruction book does not need to be grappled with right away as there is much about equallizing of the batteries etc which is not applicable to AGM batteries. The MX60 default settings (p.79) for Absorb is 28.8 V and for Float is 27.2 V. The settings recommended for an AGM battery (p.86) is the same 28.8 V for Absorb and a slightly lower 26.8 V for Float. My batteries sit in a fairly stable temperature environment, probably hovering around the 60 - 70 F mark, so I do not have a temperature sensor installed. I could have adjusted the default 28.8 V Absorb setting up a bit to make for a lower than STC temperature adjustment. Most literature warns not to overcharge AGM batteries so I left it as be. The default 27.2 V Float setting could be considered to be the 26.8 V recommended setting for AGM batteries but adjusted up 0.4 V to correspond to a temperature adjustment to 65 F (p.85). All in all the default settings were good already and so were not changed. Basically I only had to input the battery bank voltage. That was it. Starting up the Xantrex inverter was even easier as one simply sets the selectors to the right settings for battery type and Ah capacity and it is ready. Without adjusting the search mode watts potentiometer completely CCW, the inverter when turned on searches for a load that is not there but once the coffee maker is turned on it is a big load and the inverter kicks on.
Operation
Three weeks of operating data only. The batteries start each day steady at 26.2 volts, then kick down to 25.3 volts when the inverter and coffee maker are turned on, and go back to 26.2 volts shortly after both are turned off. By about 10am the batteries have gone through an MPPT stage, a 28.8 V Absorb stage and a 27.2 V Float stage and are back to fully charged. This confirms the design that it takes about 1 1/2 hours of sun to make my coffee. This also means that after 10am on a sunny day my system can and wants to put more power into my batteries but that I am not using what sunshine is available. (roughly: 200 watts panel power X 6 more hours of unused sunshine X efficiency factors = approximately 1kWh of unused power daily in September). This is good and I will have fun finding other loads to convert to solar.
Buried on page 56 of the Outback instruction manual, it turns out that the MPPT feature of the MX60 kicks in at 5 amps input. The Sanyo panel has an Isc rating of 3.83 Amps. This means that the MPPT feature will never kick in for this minimal design system. Fortunately the controller has an MPPT default setting at 77% Voc, which keeps the charge controller working in an efficient range for converting the incoming voltage down to 24V. The MX60 can handle 60 Amp input, hence there is much room for expanding the system. If I ever buy another panel to expand the system the MPPT feature would kick in.
The Xantrex DR1524 is a 'modified sine wave' inverter. It works well for powering the coffee maker heating element. Am told that some electronic equipment such as computers would need a 'true sine wave' inverter.
I am also starting to experiment with what it takes to cook with PV- my microwave is rated at 800 watts, a bit less than the coffee maker. 'Reheat' takes 3.5 minutes. Using the same formulas as above, one 'reheat' cycle of cooking with the microwave would use 2.2 Ah of power, drawing 37 amps from the batteries, which is a discharge rate of C/8.1. That is in theory. In practice I found that 'reheat' draws 12 Amps from the wall circuit (making the microwave work at a rate of 1440 watts and not the rated 800). This would use 3.9 Ah of power, drawing 67 amps from the batteries, which is a discharge rate of C/4.5. This exceeds the recommended C/5 discharge rate. Another good lesson: when doing load analysis, a short home-made extension plug to separate the individual wires and a Fluke 337 True RMS Clamp Meter for getting accurate current readings are important.
Conclusion
An unexpected result of this system is the daily enjoyment factor. Every morning I have my coffee knowing the power brewing it came from the sun. The technologies that are part of making this system work are quite amazing. Every day I make my coffee and smile. Sometimes neighbours come over and I can pour them a solar coffee. They smile too. For sure the cost was a bit high. Now that I know some of what to look for and some of the design issues, I am able to look around and see what I could have done differently, better, or cheaper. For me it was a good project. I am gaining an appreciation of load analysis and starting to understand the different loads around the house. Load awareness factor plus a daily enjoyment factor equals money well spent.
Costs 2008, Installed 2010
Parts
$ Cost
Xantrex DR1524 Inverter (1)
1100
Outback MX60 Charge Controller
700
Sanyo HIP-200BA3 PV Panel
1100
E-Panel for 250 Amp DC breaker
500
6 X ReadyWatt RW-100 Batteries (2)
600
wires
120
misc (3)
380
Total
$4500
notes:
(1) used inverter
(2) slightly used (never a good idea with batteries but was assured they were good)
(3) cable lugs, crimping, disconnects, fuses, rails/wood/cement for PV panel support