What Happens to Automation at Peak? How to Plan an Automated Warehouse for Your Busiest Weeks

Munro Footwear Group in Melbourne runs 205 Mushiny robots against 5,239 shelf pods. Alloys runs five robots against 100. Both are sized for a quiet season and a busy one.

Most sites have two volumes: a baseline, and a few weeks well above it. As an illustration, many operations plan around two to five times baseline for those weeks. Test that range against your own order history. It is not a measured Australian figure. The short version is that shelf-to-person robots and sorters flex, storage does not, and planning has to start earlier than it does for labour.

Why is peak a problem for automation at all?

A manual site solves peak with labour hire. It adds people for the weeks it needs them and lets them go. That works, but it costs a lot per head, and casuals are less practised than your regular team, so quality can slip when volume is highest.

An automated site has the opposite shape. Its capacity is bought in advance and does not leave in January. Size it for peak and it sits partly idle most of the year. Size it for average and it falls over in November. So the question is which parts you can stretch, and what you must decide before the season starts.

What scales at peak and what does not?

Do more robots mean more throughput?

Up to a point, yes. Fleet size is the variable in a shelf-to-person system. More robots means more pods reach pickers each hour, which is why Mushiny sites range from five robots to 205 on the same system. If one robot is charging or being repaired, the others keep going.

The ceiling is the grid, not the fleet. Grid robots navigate by QR markers on the floor, which tell each robot exactly where it is. A marker does not tell it where to go. Each robot still chooses its own route and negotiates with the others. This works because the robots run inside a fenced area that people do not enter. That isolation is why QR is faster, cheaper and more accurate than SLAM, the variant for shared space.

The same thing that makes the grid fast also limits it. Every robot you add shares the same aisles and the same approaches to the stations. Past some point the extra robots wait for each other and the extra throughput stops arriving. Where that point sits depends on your layout and station count, so we find it in simulation during sizing rather than quoting one number.

Do workstations scale at peak?

Within limits. The workstation count is fixed at install, and each station has a rate. A MIX 4.0 workstation presents up to 240 totes an hour, with 36 buffer locations for hot items, two queuing positions and two picking positions. More robots keep every station fed, but they cannot make a station present more than its rate, and you cannot add stations in November.

That makes the station count the real peak decision. If your peak needs more picks an hour than your stations can present, no number of extra robots helps. Count stations against your peak hour, not your average one.

Can I add racking in November?

No, and this is the part people miss. Storage is fixed once it is built. Peak usually brings more stock in before it brings more orders out, so the pressure on storage arrives ahead of the pressure on picking.

Shelf-to-person pods stand up to 2.7 metres, which adds around 20 per cent more storage on the same floor. MIX Pro stores totes double-deep in racking up to 12 metres high, with totes 20 mm apart and 900 mm aisles. Density is a design decision made long before the season. What you can flex in the season is how you use what exists. Clear slow movers before peak so seasonal lines have somewhere to go, and schedule inbound against free locations. On MIX Pro, a single aisle supports more than one gantry, so adding gantries and tote transfer robots raises how fast stock moves, though not how much the building holds.

How does sortation handle peak?

Sortation scales by hours and by configuration, not instantly. Mushiny publishes these rates.

• The Minisorter sorts 1,200 to 1,500 items an hour, with a mechanical rate of 1,800.
• The 3D Sorter MAX sorts 1,000 to 10,000 items an hour or more, with a mechanical rate of 14,400.
• Across the wider range, the Mini-D runs 1,500 to 1,800, the Pro 2,000 to 5,000 and the Multi 30,000 or more.

Destination count is the other limit, because every open order needs somewhere to land. A Minisorter holds 42 to 140 destinations for small items, 30 to 100 for medium and 24 or 48 for large. A MAX sets its count by machine length, up to 60 metres. If peak means more orders open at once than you have destinations, batches get smaller and the saving shrinks. GPA Logistics in the USA doubled their sorter from 80 destinations to 160 after it was running, but not between one week and the next.

Packing can be the next constraint. The PolyPacker bags at up to 1,500 bags an hour, and rental is available in Australia through Mushiny's partnership with Omni Group.

How do people scale at peak?

Differently, and this is the strongest argument for automation at peak. A casual in a manual aisle has to learn where things are before they are productive. In a goods-to-person system the stock comes to the station, so there are no pick faces to learn and no aisles to walk. A new starter learns the station, not the building.

Put a number on it: about half a day. That is how long a new starter takes to become productive at a goods-to-person picking station. Compare that with the weeks it takes someone to learn their way around a manual pick face. On the Minisorter the job is simple enough that casual staff over peak are not a risk for the same reason. You do not have to take our word for either. Put a new person on a station at the Chatswood Tech Centre and time them.

The limit is that people need somewhere to stand. A Minisorter has two induction ports and a 3D Sorter MAX has up to 10. Labour scales quickly, but only up to the number of places to put it.

What is the cheapest way to add capacity?

Run more hours. A system that runs one shift at baseline can run two or three at peak, which costs wages instead of capital. A MIX Pro gantry draws power from a sliding contact line, so it does not stop to charge. A MIX 4.0 battery takes 2 hours to charge and lasts 12, so a longer day needs a charging plan. It also needs supervision and maintenance cover, and Mushiny runs a 24/7 service hotline. Extra hours help only if stations are the limit, not storage.

Can robots be hired for a season?

Not at the moment. The challenge for us as a business is that most business peaks fall in November and December. If we carried a fleet of spare robots, they would sit dormant at our site for most of the year and then go to the highest bidder at peak. That makes the business case difficult, which is why we recommend designing for a reasonable volume up front and flexing through operating hours during peak.

When do I have to decide by?

Earlier than you think. Robots run five to six months from order to running on site, so a November peak is a decision made in the first half of the year, not in September. If you are reading this in spring and hoping to add robots before Christmas, that option has already closed. What is still open in September is operating hours, rostering and clearing storage, which is why the list below is about those.

What should I do in September to be ready for November?

1. Pull last year's peak, hour by hour. Find peak lines per hour, open orders and the date inbound stock started arriving.
2. Check stations and destinations against the peak hour, not the average.
3. Count free storage against inbound for the six weeks before orders peak.
4. Fix what you have been tolerating. Peak exposes whatever you did not fix in the quiet months: a replenishment lag, a slow dock, a layout that makes pickers reach.
5. Roster the extra hours and casuals early. Allow about half a day per new starter on a station.
6. Freeze changes. Do not go live on something new in October.

What is the downside?

An automated site is harder to flex upward in a hurry than a labour-hire site. If your peak arrives two weeks early or at twice what you planned, you can ring an agency tomorrow. You cannot ring anyone and have a station, a rack or a sorter destination tomorrow. The planning has to happen earlier, and some capacity will sit idle in the quiet months.

In return, your service level depends less on how many casuals you can find. Whether that trade pays depends on how big your peak is and how long your quiet months are. Bring your peak data to the Chatswood Tech Centre and see live sites on the references page.

Frequently asked questions