Cut the load before you generate it

Load removed from a site needs no roof, no connection and no inverter. It works at night, it cannot be curtailed, and it makes every system that follows smaller. That is why it comes first.

Reducing a site's energy use and generating its own energy are both worth doing. They are not worth doing in either order, and the reason is arithmetic rather than principle.

Load removed from a site stays removed in every hour it would have run, including the ones after dark. Generation is only available while the sun is up, and only worth its full value while the site is drawing.

Removed load also needs nothing built. No structural assessment, no switchboard headroom, no connection enquiry, no export limit. It does not depreciate and it cannot be curtailed.

There is a second effect that rarely makes the paper. Load reduction lowers the peak as well as the total, and the peak is what sets the connection, the switchboard headroom and often the network tariff. Removing demand can make a project possible on a board that would otherwise have needed replacing.

And it compounds. Every unit of demand taken out of a site before the array is designed reduces the array that site needs, which reduces the capital, which shortens the payback on both pieces of work at once.

The order is therefore load, then generation, then storage. Reversing it produces a system sized to a building that no longer exists by the time the efficiency work is done.

Set against that, load work is harder to sell. An array is visible, it has a capacity to quote and a photograph to show. A corrected set point and a resealed cool room door have neither.

That is a presentation problem rather than an economic one, and the fix is to report both pieces of work in the same units: kilowatt hours removed per year, and what they cost to remove.

That is the whole argument. The rest of this post is about where the load actually is, which is rarely where people expect.

The site's own interval data is the least expensive diagnostic available, and the retailer holds it. A year of it, at the finest resolution the meter records, costs a request.

Read it for base load first. Base load is what the site draws when it is doing nothing: overnight, over a weekend, across a shutdown week. It is the flattest and most honest line in the data.

A high base load is often the most valuable single finding in a commercial building, because it is demand that is not producing anything. Standby plant, lighting on circuits with no control, refrigeration running against a failed door seal, air handling that never steps down, and a compressor cycling on a leak.

Quantifying it is simple once the data is in front of you. Take the lowest sustained demand across a shutdown period, multiply by the hours in a year, and price it at the tariff. That figure is the budget case for finding out why.

It is also the demand that solar cannot touch. Base load at two in the morning is bought from the grid at whatever the tariff says, every night, for as long as it is left running.

The second reading is the daytime shape. That is what the array will be sized against, and another post owns it: see "Size it from the load, not the roof".

The third is the peak, which is the shape that costs money on a demand tariff rather than an energy one. That has its own post too: see "Demand charges are a different bill".

Base load is the one that belongs here, because it is the part the buildings themselves can be made to stop doing.

One caution on the data itself. A quarter of readings describes a quarter. A full year carries the summer peak, the winter shape and the shutdown weeks, and those are the periods that reveal what the building does when nobody is asking anything of it.

It is also the most reliable, which is why it stays at the top of it.

Lighting is a measurable load, it runs on a known schedule, and the replacement technology is mature. The result of an upgrade is close to calculable in advance, which is not true of most energy work.

It carries a second benefit that gets left out of the calculation. A luminaire that draws less also emits less heat into the space, which takes load off the air conditioning underneath it. In a high-bay warehouse or a kitchen, that secondary saving is not marginal.

The luminaire and the control are two separate savings, and a design that changes the lamp without adding control captures only one of them. A luminaire drawing less is worth less than a luminaire drawing less and switching off when the space is empty, and occupancy and daylight control are ordinary inclusions rather than extras.

The part that gets skipped is the measurement afterwards. A lighting design is a calculation of illuminance across a surface, and whether the installed result matches it is a survey question with an instrument attached. On a sports court, a warehouse aisle or a car park, the survey is the difference between a lighting upgrade and a lamp swap.

Glare and spill belong in the same conversation. Light that leaves the site is light the site paid to generate and then gave away, and on a boundary with neighbours it is also a complaint.

The replacement decision is also a maintenance decision. On a high bay, a tower or a pole, the access cost of changing a lamp exceeds the lamp several times over, and a longer service life is worth more than the wattage difference suggests.

That is the number to put in the paper next to the energy saving: how many access visits the upgrade removes across its life.

On most commercial sites, air conditioning and refrigeration are the largest controllable loads in the building, and the largest contributors to the afternoon peak.

They are also the loads most likely to be running on assumptions rather than measurements, because nobody complains about a building that is slightly too comfortable at three in the morning.

Step two deserves the emphasis it rarely gets. Controls drift. Set points get nudged during a hot week and never moved back, schedules survive a change of tenancy, and a dead band narrow enough to make the plant hunt will run it all day for no comfort benefit at all.

None of that requires capital. It requires somebody to open the controller and read what it is actually doing, against what the building is actually occupied for.

Step four is where the emissions argument and the energy argument separate. Moving a gas load to an efficient electric one raises electricity consumption and lowers total energy and combustion emissions, which looks worse on one line of a report and better on the whole of it. Say so in the paper, before somebody else notices.

Refrigeration belongs in the same review wherever a site has it. Door seals, defrost schedules, condenser cleanliness and set points are unglamorous, and together they move a load that runs every hour of the year.

By the time the load work is done, the building is a different building, and it is the right one to design an array for.

It is also a building whose peak has moved, which matters more than the total. The array is sized against the load that remains, the switchboard is assessed against the demand that remains, and both of those are smaller than they were.

The array will be smaller than the one quoted against the old consumption. That is the intended result, not a loss, and the capital not spent on panels is available for the efficiency work.

Storage comes last for the same reason it always does. A battery earns its keep on a tariff spread, on a demand charge or on the cost of an outage, and sizing it against a load that is about to fall is how sites end up with storage they never fully cycle.

The combined design question has its own post: see "Sizing solar and storage together".

One sequencing note. Where the efficiency work and the array are going in within the same year, design them together rather than in series. The array should be sized against the post-upgrade load, which means the load calculation has to exist before the layout does.

New South Wales runs certificate schemes that sit underneath much of this work, and they change the arithmetic rather than the order.

The Energy Savings Scheme creates certificates for activities that reduce electricity consumption, which is the efficiency half of this list¹. The Peak Demand Reduction Scheme creates certificates for reducing demand in the peak window, which is the storage and controls half².

Neither is a reason to do something a site does not need. Both are a reason to sequence the work so the eligible activities are captured rather than missed, and to establish eligibility before the equipment is ordered rather than after it is installed.

The sequencing point is the one that costs money when it is missed. Many activities rely on evidence of the equipment being replaced, so a site that strips out the old lighting on a Friday and asks about certificates on the Monday has usually lost the evidence.

What the certificates are worth moves, and no responsible proposal states an amount as though it were fixed. Another post covers exactly why: see "Why nobody can quote you an exact figure".

The order stays the same with or without them. Remove the load, then generate against what is left, then store only where storage has a job to do.