Give me some ideas on how to approach this one please?

General
1

Guys,

This previous thread details a little of the background - with some significant changes now.

Our power gird in Australia (NEM) is run on a 5min bidding window. I am with a wholesaler (so i get exposed to variable wholesale rates (for good and bad !)

I have a LOT of solar panels (23Kw) and a lot of Battery 40Kwh - both of which were purchased based on the ability to arbitrage the wholesale rates (which has been working well)

I want to optimise my system for maximum return now.

The way the market works is that every 5m there is a computerised bidding system that sets the price and this is published about 30sec into the interval - i have access to this in the same timeframe and can perform actions based on this price (through current automations - such as charge and discharge the battery)

Our provider publishes estimated pricing for approx 24 hours ahead for each 5 min window - the further out they are the more it is based on estimated load and supply - but obviously can not take into account any issues (power station going offline, downed power cables etc)

Typically most evenings (starting from abut 4:30PM through to about 9PM) will be the evening peak when prices are generally higher. I have a punitive demand tarriff applied to my account that means it is very expensive to pull any power from the grid between 3PM and 9PM. (take whatever my max power draw is during a 1/2 hour window in that 3-9 slot, double it and then multiply by the number of days in the month and then charge at 33c Kwh)

I currently read the current big and future forecasts into a Global Context array - there are some obvious "spikes" on a fairly frequent (every couple of days) basis and these are easily addressed.

Where i have an issue is in the edge cases and how to address them.

Lets say i need 10Kwh from my batteries to get through the night with our standard base load. This means i have approximately 30Kwh to be able to dispatch into the evening peak - however Lets say i can discharge into the grid at a 30c/Kwh return - i need to look forward through the prices and decide if there are any future forecasts that look like a better return than that 30c and also whether if i did take the 30c now - is there a cheaper opportunity after 9PM to recharge the battery to take advantage of those higher prices later (i.e. discharge, charge and then discharge again) - noting that the farther out the forecast is (in time) the less accurate it is.

My batteries has a round trip cost of approx 8c per Kwh (based on purchase price and warranty period) and this also has to be added to the charge/discharge cycle costs (i.e. it is no good chasing a 5c delta)

Does anyone have a good mathematical approach to solving this - i note that in the Home Assistant platform someone has written a Linear optimisation tool that is mean to do this and also take into account (discretionary loads and expected solar input)

Any ideas on the mathematical approach for this in NR ?

Craig

2

Surely a couple of relatively simple rules will give you 90% of the benefit from access to wholesale prices?

Always fully charge your battery during the day - ideally from solar but if that won't fill it, from the grid.
Stop importing power if the price goes above a trigger point
Always discharge to the grid during the 4 - 9pm window, leaving enough to get your property through the night.
Stop exporting if the price goes below a trigger point.

3

OT, not sure how they get away with calling it the "National" market, as it is only for the east coast of Australia.

4

Yeah unfortunately for it to be truly national - would require a LOT of very expensive wires to go a very long way (all the way from Adelaide to Perth) - which they will eventually do - but WA people see themselves as very different from the rest of OZ.

5

Yeah it sounds easy when you first glance at it - but then when you dig in is where the maths comes in

e.g. if i have an opportunity to sell at (say) 50c - and the power price for the rest of night (after 9PM) is less than 15c with a fairly high probability of being accurate - should i take it ?

If i sell at (say) 50c - and there is a possibility of selling more power later that night at a higher price - should i infact save the battery for the higher priced event later - or should i scan and look for an opportunity to recharge ?

As you say fairly basic rules for dispatch are easy and is what i have been doing up to now - but the market has just changed from 30m billing to 5m billing - so there is a lot more opportunity to work the system - hence why i was looking for some ideas of how to approach it other than just basic rules

Craig

6

Hey, watch it buddy! :rofl: :joy: