House Thermal Inertia and Roomstat Setback (some cautionary notes)

If the price at the margin is 100p and the incentive is 3p but the margin price does not change then there is not enough switching. So the market is not working properly. Unless I am missing something about the costs involved in powering up a gas-fired power station.

But then the incentive could rise to the same level as the gas fired production to show that the market has found the optimum price. I do not see why producers need high incentives to supply but consumers cannot get the same incentives to lower/shift demand.

Because most consumer will not risk having to pay more at peak times regardless of paying less at other times. Most people don’t have logical thinking skills…

1 Like

I am not saying that people should be on Agile (or other TOU tariffs). It is just an indicator of wholesale costs at that point in time. We all pay these extra costs even if we are on fixed price tariffs as they add to our provider’s costs and so have to be factored into their next price offering, whether that be a variable or fixed tariff. If the market is inefficient we all pay eventually.

Think of the news headlines if a single parant of young children claim that forced time of day pricing is preventing her cooking them hot meals…

There also allready enough people refusing smart meters as they think it a type of government control and that life would be better without net zero targets, no renewables, and maximising UK gas production.

2 Likes

Maybe you are right. This sums up our current problem. Reluctance to do the sensible thing that would benefit everyone because we fear a skewed negative headline - which will happen anyway. So the Daily Mail wins.

So do it gradually and stop if it’s not working. Maybe a 10p per kWh incentive to shift demand. It may also encourage people to get smart meters.

I’ve not verified myself, but I read on here that you can switch to LWT mode, configure a schedule of LWT adjustments, and then switch back to Madoka control. The LWT adjustments are no longer accessible, but they are still honoured.

2 Likes

But we can give reward points to people who use a higher % of their electricity at times of low demand. Maybe also a stranding charge reduction for agreeing to having smart meter.

1 Like

Thanks for the reassurance John. I will be putting my Matlab algorithm here when I get a bit of spare time.

1 Like

Ok. I’ll go with that. Push or pull, I don’t mind. Just a more efficient, fairer market that means those less well off can also benefit. And where emergency suppliers don’t game the system.

Ah, GitHub! My defence against code implosion resulting from too many untested changes! Keep pressing the commit button…

Great post @SarahH!

I can add a bit to this discussion since I have worked a lot on this topic the past year.

There is a very concise way to describe the thermal dynamics of a building using something called a lumped RC network. In literature you will find hundreds of papers that adopt this method in their work.

If you’re interested, the following papers are fantastic:

The basic idea is to model a building as a collection of thermal capacitors that store heat and thermal resistors that conduct heat. Heat can be added to the system by heat pumps, boilers, solar thermal gains, etc. Graphically that looks like this:

I added here nodes for the interior, the envelope, the emission system and ofcourse the ambient temperature. If you’ve ever worked with electronics this should look familiar to you. Whether you’re modelling voltage and current or temperature and heat flow, the equations are exactly the same!

We can find them by using the formula for conservation of energy, which means the flows going into a node must equal the flows going out of a node:

\begin{align} \sum\Phi_{\mathrm{in}} &= \sum\Phi_{\mathrm{out}} \end{align}

Where \Phi is heat flow in kW. Applying that to every temperature node gives you a system of differential equations:

\begin{align} C_\mathrm{i}dT_\mathrm{i} =& \Phi_{\mathrm{w}}(t) + \frac{T_{\mathrm{h}}(t) - T_{\mathrm{i}}(t)}{R_{\mathrm{ih}}} + \frac{T_{\mathrm{e}}(t) - T_{\mathrm{i}}(t)}{R_{\mathrm{ie}}} \\ C_\mathrm{e}dT_\mathrm{e} =& \Phi_{\mathrm{e}}(t) + \frac{T_{\mathrm{a}}(t) - T_{\mathrm{e}}(t)}{R_{\mathrm{ea}}} + \frac{T_{\mathrm{i}}(t) - T_{\mathrm{e}}(t)}{R_{\mathrm{ie}}} \\ C_\mathrm{h}dT_\mathrm{h} =& \Phi_{\mathrm{CV}}(t) + \Phi_{\mathrm{HP}}(t) + \frac{T_{\mathrm{i}}(t) - T_{\mathrm{h}}(t)}{R_{\mathrm{ih}}} \end{align}

I’ve summarized the meaning of the units in a table:

Symbol Physical Interpretation Unit
C_\mathrm{i} Thermal mass of the building interior. kWh/C
C_\mathrm{e} Thermal mass of walls, floor and envelope. kWh/C
C_\mathrm{h} Thermal mass of the heat emission system. kWh/C
R_\mathrm{ih} Thermal resistance between emission and supply. C/kW
R_\mathrm{ea} House envelope insulation or air barrier. C/kW
R_\mathrm{ie} Insulation between interior and envelope. C/kW
R_\mathrm{ia} Direct leakage from the interior to outside. C/kW

What can you do with this information? Well you can use it to answer the question whether roomstat setback is worth it or not, by directly using physics. You can also use this to build controllers, optimize your heating to variable rates, etc. For that part you also need to write more math and specify a controller, maybe I’ll write about that some other time :wink:

6 Likes

One problem here, Sarah is not an electrical engineer :worried:

Having said that, our favourite LTSpice should be able to model that very easily.

Solving these equations is essentially what LTSpice does internally. There are many other methods to do so. I myself use DifferentialEquations.jl: Efficient Differential Equation Solving in Julia · DifferentialEquations.jl.

1 Like

Of course it does, but it’s one of the quicker and easier ways to get the job done - much easier and quicker than doing it as I was taught - either mesh or nodal analysis by hand with a calculator or slide rule.

I understand. But you want these equations for two purposes: system identification and control. Neither can be done with LTSpice. Ok you can make some pretty graphs, but then we’re doing to guess work again which is what we don’t want to do :slight_smile:

If you’re writing your own control algorithm, then this is a completely different proposition to using the model to understand what’s going on and as a tool to establish or validate the ratings of the components of your system and maybe identify where changes are necessary; the first part of your answer to “What can you do with this information?”.

Yes, my point is we shouldnt have to do that and that should be the job of the heat pump.

Observing from the sideline, not having a heat pump, I very much feel that the whole heat pump supply chain is climbing the learning curve, and there’s a long way to go yet.

1 Like

Could you eloborate what you mean by that?