ok, looks good. so now if you reduce the target temperature with the ± buttons it should start to control it.
its two wire cable, low voltage so easy to move.
You can get a remote sensor if you don’t want to move the controller (MRW-TA).
1 Like
I know you don’t want to invest in a full monitoring rig, fair enough, but have you got the possibility to monitor your water temperatures during a cycle? amazon cheapo thermometers.
Hi Ian.
I have no temp recorders; all I have (in addition to the Samsung controller display (LWT and RWT) are surface-mounted temp indicators on the four buffer tank conns (4-channel indicator, no recording facility), and another pair on a living room radiator inlet/outlet. I can log these by hand, taking readings every minute or so. What’s on your mind?
Sarah
ok so you do have basic surface mounted temps and you can can tolerate the tedium of doing hand logging for one cycle every minute.
whats on the mind:
plot the graph of flow T, returnT, heat power (based on flow x DT) , elec power and roomT, every minute during your 45 min on cycle .
so you show a graph just like the ones we keep showing from emoncms, but manually.
also include thermostat satisified yes/no in the above.
I’m wondering if the reason it works for you is because you let the system get really quite cold with 45 mins off . Is it the thermostat satisfied telling it to be off at the end of the 45 mins on, and then 45 mins to cool down and back to call for heat?
After a cold startup, the time to reach "stable LWT/RWT " is usually 30-40 mins for me. I wonder if its longer for you because of the high LWT target. So its always in a “startup” phase, which is always finished by a “thermostat satisfied”. you never get a “stable phase”, or if you do its very short. But in the startup phase its not going to cycle because its constantly trying to increase LWT. So you get a 45 min runtime.
the plot of the data is to prove / disprove the above theory.
Arne already has a heat meter - does that feed into the Samsung, or is there some way to read data off of it, with modbus or something? Being able to visualise the behaviour and performance of the heatpump directly would make for much better analysis and adjustments.
Ah cool, that’s a good idea. I didn’t realise this was an option, quite cheap also. Although I guess you still need to run a cable, so might was well just move the controller
Hi @Ian_Calderbank.
No, my system also takes about 40 mins for LWT to reach WL (weather comp) setting. I’ll take a set of readings over the next few days, but I’m worried because 1) resolution of RWT is poor (only to nearest degC) and 2) the power reading seems to have large time lags which are hard to reconcile with instantaneous measured duty.
I’ll do my best and report back…
Sarah
totally agree on the need.
you can read the samsung with modbus, but only if the modbus module is installed in the samsung. which it won’t be with his system.
@glyn.hudson is there a way to read his heat meter easily with a minimal emoncms setup on a PI or something?
[Heat meter is Kamstrup, btw]
It depends, although probably unlikely unless MBUS factory fitted option was fitted
He’s got a Kamstrup 302 which needs the MBUS option selected when it was purchased. It does not have retrofitable MBUS modules like the Kamstrup 403 has. It’s probably unlikely he has this since I can’t see the factory fitted MBUS cable on the photos.
2 Likes
Hi again @Ian_Calderbank and @Timbones
I’ve taken another set of readings from my Samsung 8kW Gen 7 system, to try and understand why I don’t see the rapid cycling that a few others see, on a fairly average winter’s day (~3degC ambient). I took readings every few minutes from the ASHP controller and some surface thermocouples I’d installed on my buffer tank inlets/outlets and one radiator in the only heated room (living room), all with lots of heat transfer cement and good insulation.
Here are the raw data, along with a preliminary analysis:
OEM3.xlsx (28.9 KB)
As you will see, over an hour or so (when the circuit was fairly settled close to weather comp setpoint, and the living room was approaching roomstat setpoint of 21.5degC), the compressor was running happily at about 60% of full speed (columns AA and AB), with a couple of short (~5 min) stops during the hour, both times when the LWT (column C) reached weather comp setpoint (column U), starting again when LWT had dropped 2-3degC.
Other interesting observations (assuming for the moment that the thermocouples were reading reasonably accurately):
- Measured pipework heat losses were high. Of the ~4.8kW generated by the ASHP, ~1.6kW was lost in the primary circuit and ~1.0kW in the secondary circuit, leaving only ~2.2kW for the radiators.
- The calculated radiator duty matched the estimated heat loss almost exactly, which is odd, because the room sensible heat (~0.5degC in an hour) should have required an extra ~1.5kW (I estimate the room m.Cp to be 11000kJ/degC). I can only assume that the missing heat came from the secondary circuit heat loss (e.g. underfloor radiator piping).
I conclude that I don’t see rapid cycles mainly because (room heat loss + pipework loss) is high enough to keep the compressor speed just above turndown.
I’ll repeat the test when it’s warmer outside and room heat loss will be lower.
Sarah
1 Like
so in a way you’re lucky, the secondary side pipework’s loss is giving you underfloor heating 
is the primary side loss going into the house fabric? have you had a look at where the pipes route, what the insulation level is?
Not really @Ian_Calderbank, mostly to the garage (where the buffer tank is), plus 2 x 3m pipe run in the outside air. Pipe insulation is not high quality (about 20mm thk foam rubber).
I’d like to install a heat meter in the secondary loop (this would tell me what’s really coming into the house - my surface thermocouples may be inaccurate - and help justify pipework reinsulation), but the installation cost (maybe £500 including labour) would be hard to justify 
. (Plus I’m worried that the heat meter wouldn’t be accurate on 20% glycol,)
You would buy a heat meter that is specifically calibrated for glycol.
Or ditch the glycol and fit antifreeze valves whilst fitting the heat meter.
Or fit two heat meters and use the relative losses to work out the proportion of heat being lost to the outside world vs staying in the house; and the electricity meter to quantify the cost of this.
Or estimate it by math and knowing pipes/tanks.
I suspect the primary losses are not quite as nasty as surface thermocouples suggest (the temperature difference with heat pumps are so modest that small errors can add up to a great deal) but still nasty.
Line the garage with 70 mm thick insulated plasterboard (U-value ~0.3), throw up a plasterboard ceiling and insulated above that, add an insulated garage door, and pretend that the losses are useful as you now have a home gym / larder that doesn’t freeze / clothes cupboard that doesn’t get mildew / woodworking workshop with more stable temperature and humidity etc? 
1 Like
primary pro is easy to do yourself.
if you were going to the effort of a re-pipe, obvious suggestion is repipe with the buffer as a volumiser. then you’d only need one heat meter as well.
@SarahH @Ian_Calderbank @marko @matt-drummer @glyn.hudson
Just a little exercise in Semi Auto read out of the cycling
I thought - I can hear when it starts and stops - so a microphone would catch that as well…
So here it is - Measured over 5 hours at night on my Ipad - took the file in to Acuity - put on some compressor and amplified the signal - here it is:
First stop is longer because the pump comes out of doing DHW
I’ve drawn a couple of minor conclusions from this thread, please correct me if I’ve missed something significant.
Cycling rates can be estimated for weather compensation (WC) control of LWT and for roomstat control of ASHP operation as follows:
WC control “off” time: Emitter inventory (incl. piping/volumiser/buffer) in kg * Sp Heat in kJ/kg/degC * controller hysteresis in degC / Emitter loss in kW (incl. losses from pipework etc.).
Example: inventory = 100kg, specific heat = 4.2kJ/kg/degC, emitter energy = 3kW, WC hysteresis = 2degC. Then time to drop LWT by the hysteresis value = 100 * 4.2 * 2 / 3 = 280s (about 5min). Note - some ASHP compressors have a timed minimum off time, e.g. 5min for a Samsung Gen 7.
WC control “on” time: Emitter inventory (incl piping/volumiser/buffer) in kg * Sp Heat in kJ/kg/degC * controller hysteresis in degC / (ASHP energy - Emitter loss) in kW. Note - ASHP energy will rise with time as compressor spins up (typically takes several minutes).
Example: As above, plus assume average ASHP energy output = minimum steady state compressor output = 4kW. Then time to increase LWT by hysteresis value = 100 * 4.2 * 2 / (4 - 3) = 840s (about 14min).
WC total cycle time: In this example = 5 + 14 = 19min. [Note - the Samsung Gen 7 takes ~7mins to reach full speed - and thus energy output - from startup, so would exceed the example emitter loss considerably quicker than 14mins, probably more like half that.]
Roomstat control: Requires knowledge of the effective (m.Cp) of the room where the roomstat is located (i.e. its thermal inertia), and the effective hysteresis of the roomstat. Thermal inertia can be estimated by measuring the temperature reduction over time when the room containing the roomstat is not heated (e.g. overnight), and having a good estimate of the heat loss from this room at the known external temperature (this requires a numerical integration, as heat loss changes as the room cools, but is easy with a spreadsheet).
Wired roomstats may have directly settable hysteresis if they are simple on-off control; wireless roomstats are invariably TPI (Time Proportional Integral) controlled (“Smart” thermostats) which anticipate setpoint to minimise temperature overshoot.
Roomstat cycle “off” time: Room thermal inertia in kJ/degC * roomstat effective hysteresis in degC / room heat loss in kW.
Example: Room (m.Cp) = 10000kJ/degC, effective roomstat hysteresis = 0.5degC, room heat loss = 3kW, then time to cool room by hysteresis value = 10000 * 0.5 / 3 = 1670s (about 30mins).
Roomstat cycle “on” time: Room thermal inertia in kJ/degC * roomstat effective hysteresis in degC / (emitter output - room heat loss) in kW.
Example: As above, plus assume average emitter energy = minimum steady state compressor output = 4kW, then time to heat room by hysteresis value = 10000 * 0.5 / (4 - 3) = 5000s (about 85 mins).
Roomstat total cycle time: In this example = 30 + 85 = 115min (about 2 hours).
Of course, all these figures are dependent on your ASHP performance (e.g. min compressor speed thus duty, controller LWT hysteresis, your actual room and ambient temperatures, and any 3rd party roomstat hysteresis), but this all says to me that roomstat control (rather than weather compensation control) may result in better (i.e. reduced) cycling rates.
Counterintuitively, setting a low LWT on WC (so that WC controls the compressor, rather than the roomstat) may improve CoP performance but at the cost of increased ASHP cycling .
Effective emitter inventory. (inventory is only included at 100% if the flow rates through each radiator are exactly what they should be; in the limit case if the vast majority of water shortcuts through a towel rail of minimal volume then effective emitter inventory is just that minimal volume)
Agree.
Boiler makers would also agree with anti-cycling lockouts (setting a weather compensated minimum off time between burner cycles irrespective of what either the flow temperature or room stat based triggers are asking for)
1 Like