Here is a selection of the data collected and used; I intend to organise and collect more of it here, and graph it (and the tools to graph it). Please say if there is anything that you'd like me to do sooner...
This also contains selected circuit diagrams/schematics.
I've collected much data for our house at 16 Willingham Way (16WW) referred to in the various documents on this site.
Much of the data can be found here and is browsable, in simple text formats such as CSV, with common compression techniques applied for larger/non-live data sets.
Summary of continuing timeseries data sets available (and that may need manual archiving/attention):
To set a baseline and understand better what needed fixing, I had SAP, air-leakage and other tests done in early 2009.
Tests/results/certificates by Tophouse 2009/03/30 for us were:
Also published via OpenSensors.io using this script run every minute when generation is likely (ie less so in the winter when the off-grid energy to power it is scarce). (See approximately first year's per-minute data (.xz) to 2015/07/29.)
2008/02/25: the first phase/round of PV, 1.29kWp (west-facing, 6 x Sanyo HIT 215Wp hybrid crystalline/thinfilm panels with SMA SunnyBoy 1100 inverter), was installed.
2009/02/10: the second phase was installed, taking the system to 3.87kWp; same panels and inverters for each of the two new east-facing strings as for the first phase.
2010/04/26: the third phase (west-facing, same panels, SMA SunnyBoy 1200 inverter) of 1.29kWp was installed, taking the whole system to 5.16kWp. New tariff and generation meter for this last part.
Rob Ferber of ElectronVault wrote to me looking for intra-day generation data (with 10 minute sampling intervals), of which I had so far exposed very little because of data-cleanup issues (bizarre time offsets in the Sunny Beam apparently attempting to correct to German civil time even though I keep it on UTC, serial numbers, corrupt data, slight drift of Sunny Beam from true UTC, etc).
I capture these intraday samples manually after dusk about twice per month, and possibly on particularly 'interesting' days, of late using the SunnyBeamTool c/o Michaël Peeters on my SheevaPlug (and now, as of 2014, Raspberry Pi B+) ARM Linux system. I didn't do this continuously because I have not been able to stop the Sunny Beam apparently drawing considerable power via the USB port, and because the SheevaPlug didn't have a 'spare' USB port: I manually disconnected the USB I/O board temporarily to connect the Sunny Beam. (As of 2014/07 experimentally I have the Sunny Beam plugged in permanently, and take 5-minute samples during daylight hours.)
Rob produced lovely tweaked graphs from my raw data up to September 2012. Thanks, Rob: great job!
Prior to Rob's analysis I had made public but not announced these intraday samples from SMA's Windows application, with no explicit time correction: 2008/03/24 2008/04/01 2008/04/09 2008/04/17 2008/05/08 2008/09/13 2008/09/14 2008/09/19 2008/10/11 2008/12/13 2009/03/15 2009/04/08 2009/05/11 2009/05/24 2009/06/29
The entire raw data set is now available, including new samples as I take them, though with many caveats (beware timestamp wobbles in particular).
Here is an extract of the numbers from the PVGIS estimate not knowing anything about our local trees, etc!
Solar radiation database used: PVGIS-classic
Nominal power of the PV system: 5.2 kW (crystalline silicon)
Estimated losses due to temperature: 7.3% (using local ambient temperature)
Estimated loss due to angular reflectance effects: 4.1%
Other losses (cables, inverter etc.): 14.0%
Combined PV system losses: 23.5%
Fixed system: inclination=23°, orientation=-90° Month Ed Em Hd Hm Jan 2.78 86.0 0.73 22.6 Feb 5.36 150 1.34 37.4 Mar 8.83 274 2.16 66.8 Apr 14.40 433 3.57 107 May 17.80 552 4.51 140 Jun 18.20 545 4.65 140 Jul 18.20 564 4.69 145 Aug 15.60 483 4.02 125 Sep 10.70 322 2.72 81.7 Oct 6.57 204 1.67 51.7 Nov 3.43 103 0.89 26.8 Dec 1.94 60.1 0.52 16.2 Yearly average 10.3 315 2.63 80.0 Total for year 3780 960
Ed: Average daily electricity production from the given system (kWh)
Em: Average monthly electricity production from the given system (kWh)
Hd: Average daily sum of global irradiation per square meter received by the modules of the given system (kWh/m2)
Hm: Average sum of global irradiation per square meter received by the modules of the given system (kWh/m2)
Note also that adding just 10% (~500Wp) vertically mounted south-facing PV would add about 1/3rd to mid-winter generation from the main array (ie ~1.6kWh/d actual from the roof and ~0.5kWh/d from the new array predicted.)
Nominal power of the PV system: 0.5 kW (crystalline silicon)
Estimated losses due to temperature and low irradiance: 6.1% (using local ambient temperature)
Estimated loss due to angular reflectance effects: 4.3%
Other losses (cables, inverter etc.): 14.0%
Combined PV system losses: 22.7%
Fixed system: inclination=90°, orientation=0° Month Ed Em Hd Hm Jan 0.58 18.0 1.40 43.5 Feb 0.79 22.2 1.94 54.2 Mar 1.16 36.0 2.89 89.7 Apr 1.25 37.6 3.25 97.4 May 1.06 32.9 2.87 89.1 Jun 1.01 30.4 2.81 84.2 Jul 1.04 32.2 2.88 89.2 Aug 1.06 32.9 2.87 88.9 Sep 1.17 35.0 3.05 91.5 Oct 0.96 29.8 2.44 75.5 Nov 0.73 22.0 1.80 53.9 Dec 0.53 16.4 1.27 39.3 Yearly average 0.946 28.8 2.46 74.7 Total for year 345 896
(See the gnuplot commands for live voltage plotting.)
Partial monthly summary of local HDD12 (base temperature of 12°C) data suitable for gnuplot to digest.
See annual data too.
See 16WW water mains inlet temperatures for an idea of how much extra "lift" is required by the DHW system in winter, for example.
See a graph of local internal temperature at 16WW collected from an OpenTRV unit tethered to the Web server.
For the 16WW OpenTRV devices see:
The take-home message from the analysis is that it believes that we have our living room at the average UK 19°C and that our living room temperature varies between about 16°C and 21°C generally in a sawtooth pattern. (Note that 18°C in the morning before school seems to suffice.)
Note that from early evening 2012/12/19 I lowered the radiator flow temperature from about 60°C to about 45°C, independently of this, to see if the i30 PRC/eTRV could work with a lower heating rate. The reduced flow temperature seems to help reduce overshoot both in the target bedroom and in the living room, though the motivation was to try to test at typical heat-pump temperatures.
Note that we don't leave our house thermostat in the living room alone; we keep it lowish unless we feel cold and/or we want to force the heating on for the children's bedrooms before they go up. The latter action would be unnecessary if we had zoning with each bedroom able call for heat independently and those wasteful and soporific peaks >19°C could probably be trimmed. In the interim we should probably make sure that we get the thermostat down to ~18°C as the children go to bed to minimise overshoot.
A friend has generously provided his raw CSV data and chart too (from somewhat east of London), for comparison. In his case there is evidence of much tighter temperature control, at a similar mean level to ours, even with lower external temperatures.
At the end of 2012 I participated in a UCL Energy Institute/LoLo study on responses to thermal discomfort, in return for access to the raw data collected by them on temperature and relative humidity at 16WW, which overlaps with some of my iButton mission 4 data collection.
Three draught plots (based on the sensors at 4 heights in each room):
See the underlying raw data (note the room codes above on the filenames) in Mac .csv file format (with lines CR terminated, not CRLF or LF), c/o Stephanie Gauthier of UCL whose project this was for, here.
All historic logging data from the k8055, mainly off-grid numbers,
is available under
and a log of system changes is at the end of
the historic stats.
Here is Solar John's SOC chart reproduced with permission, along with his note:
"it provides only a rough estimate of the actual SOC":
|% Charge||Charging||At Rest||Discharging|
See also, for example Battery State-Of-Charge Chart For Voltage & Specific Gravity and Batteries Maintenance 101.
Lead-Acid Battery State of Charge vs Voltage (Richard Perez, Home Power #36, Aug/Sep 1993) notes that "At 32°F (0°C), the effect of temperature becomes pronounced enough to distinctly change not only the battery voltage vs. SOC profile, but also its useful Ampere-hour capacity. The discharge voltage curves may be depressed by as much as 0.5VDC from those shown on the graph. Charge voltages will be elevated by as much as 0.5VDC for a cold 12 Volt lead-acid battery." (This chimes with, for example, Victron's -24mV/°C threshold adjustment from a 20°C baseline.)
A few random facts that are useful for rules of thumb, etc:
Code used to collect and compute GB grid intensity numbers, live and historical, published under an BSD-style free/open licence on GitHub: reutils.
20110124: archival initial snapshot of Eclipse project:
Just for giggles, note that there's nothing secret about the makefile used to build this site, though note that for tidiness not all of the supporting scripts are in public view. If anyone really cares I can make some or all of them visible too.
The machine that serves this site is powered by local off-grid solar and wind renewable energy as far as possible, backed up by on-grid renewables including as of 2008/03 a substantial grid-tie solar PV system, and 100% renewable grid power (mainly wind) from Ecotricity; power draw is ~1.5W.
Please email corrections, comments and suggestions.
Copyright © Damon Hart-Davis 2007-2016.