Weep, for this is whats in charge of Nuclear Energy


I’ve uploaded a video of the Honorable Tina Joemat-Pettersson giving a speech at parliament.
I am strongly convinced that either she’s on drugs, or her speech writer is.

Sadly this is whats in charge of Energy in this country.

I weep for us.

Tesla Powerwall Payback Math for South Africa

The Tesla PowerWall is finally becoming available in South Africa, so I thought I’d do some “30% for math” calculations to work out payback periods.

This post is mostly inspired by the comments here – http://mybroadband.co.za/news/energy/154888-here-are-the-tesla-powerwall-systems-you-can-buy-in-south-africa.html

Thanks to MacAfrican for the criticism in my comments, otherwise I’d be too lazy to do this post 🙂

Onto the math –

Predicted cost at the moment for a daily use PowerWall is $4000.
Rand is currently hovering at R16 (lets hope Zuma doesn’t open his mouth in the near future, as that historically has lead to large drops in Rand values).

Assuming R16 x 4000$, we have a cost of R64,000 for the battery itself.
Obviously thats a large sum of money.

Does it make sense / cents to buy one?

Lets have a look. First up we need to try to work out total lifetime costs.

The warranty for the Tesla on the NaturalSolar.com.au site indicates this

The Tesla Powerwall includes a 10 year limited warranty.
The limited warranty covers defects in parts and workmanship, as well as at least 60% energy retention after 10 years, provided it is registered and used as intended.
The Tesla Powerwall is designed for daily use applications like self-consumption of solar and load shifting. Assuming full daily cycles, Tesla Powerwall is designed to provide energy for 3650 full equivalent cycles which is equivalent to 10 years of use.

Thats interesting, as it now gives us an indication of cycle usage.
From that, I can infer that each year we’ll see a drop of around 5% in capacity.

So year 2, we’ll see 95% of original capacity, year 10 down to 60% of original capacity, and at say Year 15, around 30% of original capacity. At year 15, I’d probably want to replace the unit, or have it as a secondary storage device..

With that in mind, we can do some math!

I’ve made a basic spreadsheet using those figures and worked out payback periods for the units.

I can’t predict Eskom pricing, so I’ve gone with current CoCT pricing per KW, and worked with annual % increase’s.
Total lifetime I’ve kept to 15 years, although you could probably scrape another year or two out of the units. I expect battery replacements to at least have halved in current Rand / Dollar terms in 10 years though, so replacement should be cheaper assuming Zuma doesn’t do any more Nene’s..

Below is what it looks like for a 5% annual increase

Screen Shot 2016-02-15 at 11.50.25 AM

You’ll see that it currently doesn’t make sense to use a PowerWall at a yearly 5% increase, even at a 15 year time frame. It comes close, but no cigar..

What happens at 10%?

10% increase

At a 10% annual increase (which might be closer to what Eskom pricing will eventually be than at 5%), we see breakeven in the 12th year of ownership. By 15 years we’re safely into profit.

Lets look at a best case – well, “worst case” scenario with a yearly 15% increase:

15% increase

15% annual increase see’s break even at Year 10.

Its unlikely that we’ll see continued 15% increases though, I guesstimate using thumbsuck that we’ll see continual annual increases of 8%, which leaves us breaking even at around Year 13.

Feel free to play around with the values, I’ve uploaded the Numbers file here (as I’m a larney Mac user), or as an Excel sheet here.

In other news, am fully expecting Rich from HomeBug to critique this, hehe 😉

Some points to note:

The Rand Dollar rate is going to be the main cost influence on whether the PowerWall makes sense. If the rand drops further (and the indications are that it will), then it doesn’t make sense at R20/ dollar. If by some miracle the rand recovers to say R14 or R12 to the dollar, buying a PowerWall is a no-brainer.

NERSA approved increases may or may not beat my guesstimates. Historically we’re much more expensive per KW than 10 years ago by a large factor, so its likely that a moderate value of 10% increase per annum is going to correlate with actual figures. This will also increase once Eskom/ Muni’s introduce further daily connection fee’s and other non tariff increases on top of per KW pricing.
(Actual historical figures can be found here – http://www.eskom.co.za/CustomerCare/TariffsAndCharges/Pages/Tariff_History.aspx )

I don’t calculate round trip values for Electricity in /out of the PowerWall. Tesla documentation indicates that this is 92%, so final KW generation figures probably should be discounted by 8% for further accuracy.

I also assume you’ll be generating electricity to go into the unit from a solar install. Costs for that are not included, as we are looking purely at the viability of the PowerWall. While I can do full system calculations, its already clear that Solar generation is already cheaper than Eskom in South Africa, and has been for a few years no. Rehashing that again is of no interest to me.

Being efficient – Roof insulation – keep it cool in summer, and hot in winter!

Energy efficiency is something everyone can aim for – its typically only a minor expense and long term you save. Heck, given our electricity costs, short term we save!

Today, I look at roof insulation.

In Cape Town, we have hot summers, and coldish winters. Sure, it *feels* cold to us, but I’ve spent many a winter in far colder places.

The worst being Harbin at -40c!
I can safely say:
#1 it does freeze before it hits the ground.
#2 don’t eat the yellow snow 🙂

Back to warmer climes though –

SANS Efficiency rules (SANS 10400-XA) recommends an R value of 3.7 for insulation in Cape Town homes. (This is over and above the typical R values for roofing).

See below graphic for an indication of R values (required insulation for SA)
climate zones sa

Whats an R value, I hear you ask?
An R value is the capacity of an insulating material to resist heat flow.
The R is for… resistance.

On average:
A roof typically has an R value of 30.
A brick wall an R value of 19.

The higher the R value, the less you lose to heat.
Obviously in winter, we want to ensure that the R values for the house are high, so that we don’t lose any heat. Adding insulation to improve the roof R values is a no brainer.

What about summer?
As everyone knows, heat rises.
Surely insulating the roof prevents that heat from escaping?

Yes… However, what most people don’t think about is that although heat rises, the entire roof is also a heat sink.

Go into the roof in on a summer day. How much hotter is it than the rest of the house? Heat rises, but most of the heat is from the sun to roof, not rooms to ceiling.

Think of it this way –

The sun is above.
The roof is above.

Most roofs are made from metal or tile, those absorb a lot of heat.

The thing thats going to get heated the most is the roof area.
Insulating that will prevent a lot of the heat coming in from that area.

Heat does rise, but the majority of the additional heat in the home is generated from the roof in summer, not the rest of the house.

So, what to do?
That same insulation that helps us keep the house warm in winter helps us in summer. Insulating the roof prevents the heat from the roof area getting back into the house.

But…. what about the heat in the roof from winter? – don’t we want that?
Yes, we do, but keeping the rooms to a stable temperature is more important.
We live in the rooms, not the roof, so we want to keep the rooms as stable a temperature as possible. Insulating them top (ceiling+insulation), bottom (floor), and sides (walls) makes that happen.

Ok, so getting insulation makes sense for both winter, and summer. What should I buy?
Going back to the SANS values per region, we have a recommended number for insulation values.
Note: That number is a *minimum* not a recommendation. Your best bet is to always exceed the minimum where it makes financial sense.

So, again Cape Town has an recommended R value of 3.7 for roof insulation.
What that means in people terms is that for effective cooling / heating, Isotherm or similar insulation needs to be about 135MM or so thicker to exceed that value.

What insulation should I get?
There are a couple of readily available options:

Isotherm (Plastic based)
My personal choice is to recommend Isotherm for insulation. Its basically plastic, so is harmless to work with, and it won’t degrade like other products. Its really easy to work with too. It doesn’t have issues with water either, so is less of an issue with leaks. DIY’ing a roof install with it is fast and easy too. The rolls are light and it pretty much lays itself out when you unroll it. Once again, remember that our Cape Town R value is 3.7, so a minimum of 135mm is recommended. Again, if you can afford thicker rolls do it.

Think Pink (Fibreglass based)
While this offers marginally higher R values than Isotherm, its a pain to work with.
If you’re getting a professional installer to do it, then this is your choice, if you’re DIY’ing – avoid. Needs face masks, and gloves, and makes working in the ceiling space annoying afterwards as the fibreglass gets into your skin. Downsides – its a little worse than Isotherm with regards to water, but nowhere near as bad as cellulose.
Again, recommended thickness is at least 135mm for Cape Town.

Blown paper/ cellulose
(Just say no)
This is usually the cheapest solution. Usually its blown into the air space with a blower and the benefits are that its extremely fast to install. You need to put down at least 3-4 inches though. There are major downsides – while it offers moderate R values initially, it does tend to settle, and the R values reduce somewhat after a few months. It also has bad issues if there is a leak, as it absorbs the water, and the resulting weight can collapse the ceiling boards. Lots of examples of that on google images.
Lastly it also gets everywhere – you’ll be finding bits floating down for years to come.

Other fun solutions
Paint the roof white, as this is the most reflective color.

Mount solar panels, and kill 2 birds with one stone – my roof is partially covered by solar panels, this helps, as they’re mounted above the roof, and keep the underlying area cool, plus generate electricity. Its a little more expensive than roof insulation though!

Install those roof mounted twirly extraction fans, as those help drawing the hot air out.

Further reading:
http://www.nrel.gov/docs/legosti/old/15771.pdf – Rather good US Gov Document on this topic.


I’ve “almost” completed my house in Noordhoek, and its completely offgrid power-wise.

Currently I have 6 x 300W panels, 2 x 220Ah GEL Lead Acid batteries, an FM80, and a Victron 3KW Multiplus running, which gives me about 2KW of usable battery +- in 16 hr period, or in easier to understand terms – about 125W / hr of usage outside of daytime.

Panels come in from the roof, in 2 strings of 3 panels each.
The FM80 can support 150V / 64A input max, so I have to put 3 panels max per string for safety reasons.
36V x 3 @ 8A x2 / 108V /8A per string.
For a 108V / 16A into the FM80.

This then comes out of the FM80 to a 24V battery bank.
The FM80 charges the in the early mornings, and keeps it topped up during the day if necessary.
The 24V battery bank is connected to a Victron Multiplus 3000/24, which then provides AC out to the house.

Batteries are 24v / 220Ah (5280W total capacity). 12v / 220Ah in series to double the voltage to 24v.
Usable capacity is 2600W overnight assuming a 50% discharge.
Overnight I actually use less than 1KW or so, so I discharge to about 80% overnight from a full battery.
Thats a 20% daily DoD, which should give me a decent lifetime.

This isn’t much!
Obviously the limiting factor here is storage (batteries), as the panels can generate far more than is needed to recharge the batteries or run the house.

What does my house need anyway in terms of power?

Lighting is all filament LED, so is minimal
Kitchen = 5 x 3W
Bedroom = 1 x 8W
Bathroom = 3 x 3W

32W inside with all the interior lighting turned on.

Hot water is provided by a solar geyser.
I’m not planning on supplementing the hot water with electricity to heat in winter unless I need to shunt excess power out from the solar panels at the inverter, so thats zero usage.

Despite the complete lack of information on most websites regarding power usage for Fridge / Freezers, I’ve decided on this model – Bosch KGW36XL30

It claims to use 227 KWhr/year, so thats about 620W /day (or 25W / hr), which is quite reasonable.

Its R7,000 though, so not cheap!

Useful site for checking KW ratings here – http://www.shopmania.co.za, as most of the manufacturer sites are useless.

Gas oven / hob setup, so zero power usage (SMEG 60cm unit).
Electric Kettles, and toasters are a no-no, as they suck too much juice.

Washing machine should use about 1-2KW a wash, so can only be run daytime when there is sun.

LED TV should be fine, as they’re about 50-70W on average.

Not much headroom for other electronics, but its sufficient for the basics. When I make more money, I can ship over the Lithium batteries 🙂

Tesla’s Powerwall

As there is a lot of interest on Tesla’s Powerwall, I thought I’d collate what I could find out about it.

The official press release details are here – http://www.teslamotors.com/presskit/teslaenergy and the end user details here http://www.teslamotors.com/powerwall


From those pages we can glean this:

Mounting: Wall Mounted Indoor/Outdoor
Inverter: Pairs with growing list of inverters
Energy: 7 kWh or 10 kWh
Continuous Power: 2 kW
Peak Power: 3.3 kW
Voltage: 350 – 450 volts
Current: 5.8 amp nominal, 8.6 amp peak output
Round Trip Efficiency: >92%
Operating Temperature Range: -20C (-4F) to 43C (110F)
Warranty: 10 years
Weight: 100kg
Dimensions: H: 1300mm W: 860mm D:180mm

There are 2 different Powerwall units available:

10KWhr for weekly cycle use @ 3500$ to installers.
7KWhr for daily cycle use @ 3000$ to installers.

Up to 9 units can be chained together to provide more power.

Simple math for the given values shows that the continuous / peak sort of make sense –
350V @ 5.8A = 2030W +-
350V @ 8.6A = 3010W +-

The voltage is curious though – at that sort of voltage, it means you will be connecting to a hybrid grid tie inverter rather than a battery inverter. Most commercial battery inverters run at 24/36/48 (or other higher multiples of 12), so that eliminates using those (if the voltages are correct).

The press pages do mention that the Powerwall is supported by the Fronius Symo Hybrid inverter.
The datasheet for those is here – http://www.fronius.com/cps/rde/xbcr/SID-3DAF0C16-CD8F5048/fronius_international/SE_DS_Fronius_Energy_package_EN_386411_snapshot.pdf

If we take a look at that, we can see that they only seem to offer a 3 phase solution currently, and that it talks to batteries using Modbus RTU, and at various voltages, depending on capacity.

Fronius’s 10.5KWh unit offers 8.4KWh capacity (80% DoD), and connects at between 280-400V @ 16A, so would be the closest offering they have to the Tesla Powerwall. Its also about twice the price!

There is other data out there though. This page purports to show details of TESLA’s 10KWhr unit.
(pictured below)

Screen Shot 2015-05-04 at 5.36.30 AM

That unit (if its real) looks a lot more compatible with battery chargers as it runs at a more common 48v.

What batteries are they using?
Tesla are using NCA batteries (LiNiCoAl)- Lithium Nickel Cobalt Aluminum in the Powerwall 10KW (weekly cycle unit)

Tesla are using NMC (LiNiMnCo) in the 7KW unit.

Elon Musk – quoted from the recent press conference:

The 10kWh device is designed as back-up, suitable for 60-70 cycles per year. Its chemistry is similar to the Tesla Model S electric vehicle, and is nickel-cobalt-aluminium cathode.

The 7kWh system is designed for daily cycling – when homes and businesses will store solar electricity produced during the day. Its daily cycling control constituent is nickel-manganese-cobalt, and Musk expects it to daily cycle for “something on the order” of 15 years.

“Actually the warranty period would be a little bit less than that,” Musk said.

“But we expect it to be something that’s in the kind of 5,000 cycle range capability, whereas the high-energy pack is more like around the maybe depending upon on how it’s used anywhere from 1,000 cycles to 1,500 cycles. And they have comparable calendar lives, and for the high energy one, it’s important to appreciate that this actually has a lot of interest from utilities “

What actual capacity is the 7KWhr unit? (daily cycle)

 Option 1

The 7KW unit is really a 10KW unit, and they’re running it at 70% of actual capacity.


Lifetime for NCA is in the 10 year range (3000 cycles +-) @ 70% DoD (depth of discharge) before it drops below 80% of original capacity.

Lithium doesn’t like being 100% charged, and it doesn’t like being 100% discharged.
Data sheets indicate that Lithium prefers 15-85% for longer lifetimes, so thats the “sweet spot”.

 Option 2
It might be that the 7KW unit is only 7KW storage though, which means actual capacity is really 70% of that or just under 5KW (4.9KWhr usable).

I personally think that its going to be option 2, sadly.

Why the cooling?
Lithium and other battery lifetime is extended dramatically if you keep it cool.
I’ll assume for lifetime purposes they’ll try to keep the batteries down to 25C

Why is it a game changer?
Pricing is about half of current retail battery units.

…but Lead Acid is cheaper!
Lead Acid isn’t cheaper.

Lead Acid provides about 30% usable capacity for a battery.
Lithium provides about 70% usable capacity for a battery.

So, if you need 10KW of usable storage, you need
30KW of Lead Acid
14KW of Lithium

Lead Acid will take up far more space.
Have less lifetime.

Whats the difference between the 10KW unit and the 7KW unit?
Other than the sizing / cycle usage, we don’t know (yet).

10KW unit is rated for weekly cycle, so would probably be for UPS style usage – eg to assist with grid outages like Eskom’s regular ones. Given Eskom’s reliability though, the 7KW unit may be more applicable here!

7KW unit is rated for daily cycle, so would probably lend itself to off grid, or time shifting, or if power outages get to be regular day to day outages here; ideal for that.
South Africa doesn’t currently have time of day rates for residential end users, although thats coming at some point, so time shifting isn’t useful here (yet).

Some thoughts on this –
If the 10KW unit is rated for weekly discharge @ 10 years (3650 / 7), thats gives you roughly 500 cycles.
500 cycles for NCA batteries @ 100% DoD is the rough lifetime, so it looks likely that 10KW is actually 10KW.

If the 7KW unit is rated for daily discharge @ 10 years, thats 3650 odd cycles. 70% DoD for NCA batteries gives about 3500 cycles before it hits 80% of original total capacity, so would indicate that the 7KW unit probably is going to be 4.9KW usable, as that also matches with the lifetime / usage.

Their pricing points for the 2 units sort of confirms it – 10KW @ 3500$, 7KW @ 3000$

So, my guess is that the 7KW is going to turn out to be 4.9KW usable out of 7KW actual storage.

Some real life testing for the NCA battery tech Tesla uses here – http://ma.ecsdl.org/content/MA2011-02/17/1282.full.pdf+html

What do I need to use a Powerwall?
At a best guess, it looks like its intended to be used with a hybrid inverter.
A hybrid inverter is a grid tied inverter with battery charging capability that can create a “mini house grid” if Eskom goes off.

So far Tesla says it works with the Fronius Symo Hybrid inverter.

Assuming that there really are a number of different units, it looks like they may offer a 48V version for those with existing MPPT chargers (as per the graphic up above somewhere).

Fronius’s Symo Hybrid inverter uses Modbus RTU (over RS485 – serial) to talk to its batteries, so I would expect the Powerwall to support that (amongst other things)

The (alleged) data sheet says IEC 61850 ( http://en.wikipedia.org/wiki/IEC_61850 )

Tesla has also been advertising for staff who have experience with RS485, Modbus, IEC 61850 and DNP3

IEC 61850 is the european favoured smart grid standard, DNP3 is the american standard.

Support for those is more likely to be in place for Tesla’s larger units eg the 100MW and up options.

Whats the price likely to be in South Africa?
If one counts in duties, clearance, shipping it will probably be in the 5000$ range or R60,000 for the 7KW.

I’ll try to keep this post updated when more info comes out.
Feel free to ask questions in the comments.

Eskom / Municipality vs Solar pricing Maths.

One of the age old questions I get asked is this – Does solar make cents(sic)?
The smartass answer is of course “it depends”.

Eskom is fast turning that answer into “extremely well” though.

Looking at the math, the average household with say 1100KW usage a month or R1800 a month average bill in Cape Town pays these rates:

600KW @ 1.56 = R936
500KW @ 1.86 = R930 (over 600KW is billed at higher rates)

Monthly thats R1866. Lets round that down to R1800 for ease of use.

1100KW / month is 36KW / day.

36KW daily usage = 1.5KW/hr on average. Ouch.

That’s quite high. Lets bring that down. We did.

First steps

Install solar hot water heating for hot water (and pool heating if you have a pool).

That should bring our bill down about 40-50%, as heating water is a major consumer of electricity.
Install gas for cooking. (We didn’t as we don’t cook that often, and it didn’t make sense in our situation)
Install LED lighting instead of power sucking halogens and regular bulbs.
We should be looking at closer to 20KW day now.

You’ll probably have spent up to R30,000 on that.


Lesson #1
Its ALWAYS cheaper to first reduce costs before going solar.
Our best bang per buck is *always* to reduce our monthly usage first.

In our case, we installed 2 x 150L solar hot water heaters.
Replaced *all* the lighting with LED’s. Don’t forget the outdoor security lighting!
Our gate light was on 12hrs a day. We replaced it with 10W LED lighting.

Outdoor security lighting @ 10W /hr @ 12hrs vs Halogen @ 150W @ 12hrs= *major savings*.
eg 120W usage for 12hrs, vs 1800W!
One 10W LED run for 12Hrs a day = 120W = (monthly 30 x 12Hrs) = R5.4
One 150W Halogen run for 12Hrs a day = R2.7 a *day*, or R81 a month.

Imagine that for all the lights in the house. If you have Halogen downlights get rid of them, and get LED ones. Takes less than a month *per* bulb for payback time…

After all that, our electricity usage went down from 1100KW/ month to about 600KW month.
That’s a 3 year payback on investment if it’s similar to our R30,000 cost.

Sure, but thats got nothing to do with Solar I hear you say.
Well, yes it does. Again, *reduce your footprint* first.

Second steps

Install some PV!

180k will get the average house with 20kw daily usage offgrid including batteries in todays money.
(Say about 5kw panels on the roof, and 30kw of battery, plus a 3kw backup generator to cater for repeated winter outages past 2 days of no sun, and all inverters etc for a single phase household)

Some Math / Justification on that
20KW daily usage = 830W/hr on average.
5KW panels will generate over 15KW in winter, and well over 30KW in summer daily, so deficit is 5KW/day or zero in summer.

Assuming 5KW / day worst case scenario deficit
You need roughly 3 x (3 days of battery) x 2 (50% discharge) for usage.
– Batteries shouldn’t be drained past 50%, so halve the rated value.
– Cater for 3 days of worst case scenario of no sun.
– Add a generator for generation for worst case scenario getting worse, and batteries go below that point of discharge.

With that in mind, deficit is 5KW odd in winter, so 3 x 5 = 15KW for 3 days of discharge (say 3 days of cloudy weather) x 2 (can’t discharge lead acid/agm/crystal batteries more than 50%) = roughly 30KW required in batteries.

28.8KW of battery can be had for a little over R1/Whr eg / 20x12v@120Ah= 28800W which can be run in 24V or 48V easily (battery inverters usually run in 24v or 48v sizing)

12v@120Ah Gel Lead Acid is currently R1500 at retail, or less, which = R30,000 for 28.8KW per 5 years usage worse case.
The good news is that battery prices are headed down, not up.

5KW of panels looks like 18 panels * 300W
300W panels are in the R11/w range retail, so roughly R60,000

Panels + Batteries = R90,000

MPPT PV Inverter should be about R20,000 (or less)
Mounting + 3KW Generator say R10,000
DC -> AC Battery inverter about R20,000 (or less), oversized so that the system is scalable if necessary.

Total so far – R140,000
Add installation, say another R10,000 (1-2 days of work) and replacement batteries in 5 years, and you’re looking at closer to R180,000 for an offgrid system over a 10 year lifespan.

If you start looking at that over 10 year terms, that’s a lot more affordable, even if you cater in replacing batteries every 5 years.

Our monthly bill is only R1000 a month though at 600KW usage / month. (Usage of R936 + other costs), and R180,000 is closer to R1500 a month. There’s a big discrepancy.

R1000 a month for 600KW x 12 = R12,000
R12,000 x 10 = R 120,000.

Our costs are closer to R180,000.

Sure. It doesn’t make sense. Its 50% more expensive!
However, that’s at todays pricing.

Nersa has granted Eskom a 15%+ increase (and Eskom is asking for more now, as the situation is dire).
15% increase on Eskom pricing means that Year 2 monthly rates are now R1150 for our 20KW/day usage.
The municipality is likely going to add a few % on top of that also, as they’ve asked for 7% also (also to be confirmed).

So year one is R12,000/ R1000 monthly
Year two is R13,800 / R1150 monthly (15% increase, which is looking lower than the actual increase will be).
Year three is R15,180 (assuming a meagre 10% increase on year 2), and so on..

Guess what just happened – our costs have (not so magically) equalised with our investment, and fairly quickly at that. Without guessing whats going to happen for year four to year ten, its already looking like a smart decision to have gone solar. We also have a nice equity in a system that has increased our house value, AND we have a system thats more reliable than Eskom is.

I know we’re happy paying a premium for the first year or two just to have electricity 24/7.

Essentially, if you have a R1500 bill a month in electricity now, and you have the capital, and roof space for it, its roughly time to start looking at going completely offgrid, as it will payoff by the 10 year mark.

I’m happy paying that premium to have reliable electricity in house right now, and I’ll guarantee you that the costs will be cheaper for self generation based than you are billed for electricity within 5 years.

Footnote –
You’ll note that I haven’t looked at feeding back to the grid in the above scenario.


It doesn’t make financial sense (at least for Cape Town). I’ll leave it up to the reader to discover why, and do the math (or look at the comments on previous articles where someone did the math!).

Getting into Hot water (underwater heating systems)

I’ve just bought a plot of land over near Capri (De Oude Weg), and intend to build a house on it.

One of the things I’d like to do is install an underwater heating system (hydronic heating) for the house. I almost installed one in my apartment in Shanghai a few years ago, and regret not doing it, especially when I visited friends who had done it!

Underfloor heating has literally been around for thousands of years, so its not a new option.
As the romans and greeks noted, heating is best situated at the ground level, as heat rises, and the area’s that we live in will stay warm.


So, thats what I intend to implement.

While I don’t even have foundations in place, or even architect’s plans, let alone any council bits done, I have already been looking at getting the underfloor bits ready, so I can order and ship a container over. Priorities!

Underfloor heating is best planned and implemented before/during a house build, so now is the perfect time for it from my perspective.

Underfloor heating essentially means running piping in a loop under the flooring.
The piping comes back to a central location to a manifold, and hot water (or cold water if you want to cool a house, eg in summer) is piped through the manifold to each room.

NB – Its better to run parallel loops to increase heat efficiency, rather than a long in / out loop.


This piping can be laid directly into the foundation slab, or on top of it, usually with insulation above the slab.

There are pro’s and con’s to both methods.

With the piping in the foundation, the concrete gets heated, and keeps its heat for a long period. The concrete holds the heat and gives it out much like a stone holds the heat in the sun.

European installs typically do underfloor heating
with a 50mm overlay concrete with insulation between the overlay and the slab (insulation above the slab).

This has three major advantages:

  • Repair or replacement of the pipes is easier. With the pipes in the slab replacement is pretty much impossible.
  • The floor temperature can be easier adjusted to changing conditions and less heat loss.
  • Other services can be accommodated in the floor.



Interestingly its important to insulate the sides of the floor also, more heat escapes that way than up, as heat escapes to where the resistance is least.
This can be mitigated somewhat by laying the piping away from the exterior walls.

I’m probably going to use ICF (Insulated Concrete Formwork) for the outer foundation walls, with a concrete mix of 20 mpa, 100 slump, 13mm aggregate for the foundation and walls.

This is where the walls have an inner and outer styrofoam layer and the concrete is poured inside.

It provides great insulation and sound proofing, and speeds up build, plus its greener. Win/win!

I’ll then lay the pipe in the foundation or above a insulation layer above the foundation (as per the diagram above) in a parallel loop using PEX conduit routed to a central location.

The conduit will be held in place using clips –
pipe clips

or using a mesh


Each room will have a different pipe conduit (probably using 100M lengths) back to the central location (aka maintenance closet), this will allow me to use 20cm spacing (works out to roughly 5M of pipe per M2 of room).

Some calculations are available here for those interested – http://www.heatweb.com/techtips/Underfloor/underfloorheating.html

I’ll update more when I have shopped more!

Going off-grid with a grid-tied inverter aka a Hybrid System

As Eskom has again gone down the load shedding route, and it looks like there won’t be much respite until Medupi et al come online, I thought I’d look at how to take my grid-tied setup to the next level, and incorporate off-grid + batteries so that we have power when the rest of the suburb does not.

I haven’t done that so far, as thats by far the most expensive part of going off-grid!

A flash file explaining a possible setup below:

There are a number of issues to deal with first.

Currently we have a grid-tied PV inverter. This expects power to be present before it supplies power from the solar panels to our house (G83 / G59 standards as per NERSA specs etc). When power goes out from Eskom, then the inverter cuts off as well (by design). What that effectively means, is that when Eskom goes off, we go off, even though daytime we could still be running off the panels.

In order to prevent that happening, I need a battery inverter system in front of the grid-tied inverter, that can automatically switch over to battery when Eskom goes offline.
Then, when Eskom goes offline, it kicks in immediately to provide power, and the grid-tied inverter doesn’t go off. Sounds simple enough, but there are a few wrinkles.

Eskom obviously doesn’t want you feeding back into their grid when its off. With the above setup, there is nothing stopping the battery inverter system from feeding back out . Thats a no-no!
This is important to follow for 2 reasons.

First, and most importantly – safety. If a lineman is doing maintenance, its better not to try to electrocute them!
Second, you don’t want your house setup to be attempting to supply the neighbourhood with power when Eskom is offline, as its quickly going to run out!

So, we need another device to sit in front of the meter that will turn off a relay to stop power going outbound if Eskom has no power, and turn it back on when Eskom is back online to allow it again. Those devices are typically called ATS’s (Automatic Transfer Switches), and are useful to prevent anti-islanding (feeding back into the grid when the grid is off).

To further complicate things, I have a 3 phase system.
Luckily ATS’s come in 3 phase varieties, and aren’t super expensive.

Having 3 phase means I need to consider whether I want a full 3 phase setup, or I can wing it with a battery inverter setup on 1 phase only to provide emergency power.

If I go the 3 phase route, then I can charge the system from solar even when Eskom is down, and I’ll have longer runtimes. This is obviously the best choice from a provisioning perspective. Costwise – not so much!

There is yet another issue though (aren’t there always!) – the battery inverter needs to be able to stop charging the batteries when they’re full.
If Eskom is off, and the battery inverter is running, and its daytime, our PV excess will be charging the battery / inverter. That needs to turn off charging to the batteries somehow when full, so that they don’t spontaneously combust. Some solutions do things like changing the frequency to the PV inverter so that it shuts off. The battery inverter then solely runs the house, and then that drains the batteries till they need more charge, and the battery inverter syncs again to provide power to the PV inverter, so that it kicks in and starts generating, and this lovely inefficient cycle of retardedness rinse / repeats ad nauseum stressing the system. I’d like to avoid that sort of solution, as it makes me, and the equipment unhappy.

That issue is described in much better detail here.

If I go single phase, then I can’t charge the system from solar when Eskom is down as the PV inverter will also be off, but I can keep one phase alive and running from batteries until Eskom deigns to do its job.
Single phase diagram (assume my inverter is 3 phase, image is snarfed from here)

off grid victron with ATS3pc0125

Choices Choices Choices…

Obviously I would like to go the 3 phase route, but that means 3 x the cost in inverters, as there don’t seem to be 3 phase inverter chargers at the consumer level in the market yet! (pocket goes ouch, and thats before I even look at batteries!).

Those who’ve read this far, may be wondering why I didn’t get a hybrid grid-tied inverter / charger in the first place, instead of sticking more stuff in front of my existing setup.
2 main reasons.
Cost, and uh cost.

Seriously – the 3 phase hybrid solutions are in the R400K and up range!

Now I’ve explained some of the pitfalls, I’ll look at some of the solutions (to be expanded as I continue my research)

ATS Switches
Comap Mainspro. $$ – This comes in a 3 phase and single phase switch setup, and can switch over automatically to prevent outage. I’d use one of these to prevent Eskom getting power, and also to switch inputs when the power goes off before any of our devices noticed there was an issue

ATS4BC0100. $ – This comes in a 3 phase switch setup (or single), and can be programmed to switch over as appropriate to inverter or mains to remove Eskom similar to the Comap.

GIS Control Module. ? – Fully automated integration of any Grid Tie Inverter System with any Battery Inverter System.Automated Load Dumping and Load Isolation based on battery charge state and solar array gain. Battery charge level monitoring and protection.

Charger Inverters
Magnum’s MS-PAE does AC coupled grid tie, but they do the aforementioned sillyness in turning off the PV inverter when their battery setup is full.
Other options.
Victron Quattro
Studer Extender series +

Reading material:
Anti-blackout system for grid connected solar installations (Solsafe / Studer)

Discussion of someones install (single phase)