In pictures: robotic milking on grass-based system

Much is being made of the shortage of labour to work in the Irish dairy industry. The headline figures of potentially 6,000 new jobs on dairy farms by 2025, while hugely positive, is a challenge to the sector.

Attracting people to work on dairy farms has always been a challenge, but with the economy at nearly full employment and lots of opportunities in other sectors, enticing people to work on dairy farms is becoming more challenging. But can technology provide a solution?

Robots

The first automatic milking system was installed in the Kilworth farm at Teagasc Moorepark in 2013 and that study with a Fullwood robot lasted until 2015. In late 2016, a new Lely robot was installed. This is the first full year of the new study. The Irish Farmers Journal visited the farm last week.

Caroline O’Sullivan is the technician working on the study. She is in charge of the day-to-day management of the cows, grass and robot. At the moment, there are 84 cows being milked. There were 86 cows trained to the robot at the start but two cows were culled for health reasons. All cows were able to be trained to the robot.

Of the 86 cows, 75 are Holstein Friesian, 10 are Jersey crossbred and there’s one Norwegian Red. Twenty seven percent are first lactation, 20% are second lactation and the rest are mature cows. There are 28.2ha assigned to the robot, so the stocking rate is 2.98 cows/ha. The robot is situated more or less in the centre point of the farm.

The big difference between this robotic milking system and that used in other farms and in other countries is the number of cows per robot. In many high input systems, there will be around 50 cows per robot and the cows will be milked between two and three times per day.

But in the Moorepark study, there are nearly 90 cows being milked. The objective of the study is not to maximise output per cow (by milking them more often) but to maximise output per robot.

Teagasc researcher Bernie O’Brien says that to achieve high milk output per robot and to avoid very low milking frequency per cow the focus of the study is on finding an appropriate grazing strategy, managing milking frequency at peak and making sure cow flow is as efficient as possible.

The last trial looked at ABC grazing, but this trial is looking at ABCD grazing using four grazing blocks. The grazing times for each blocks are as follows:

That is not to say that all cows move into and out of the blocks at these times, it just means that as cows exit the yard, they will be sent to these blocks. When I was on the farm just before midday, there were cows in A, B and C and before long they were going to start moving to D.

“Blocks A and D are at opposite ends of the farm and are always grazed in the daytime so we give them a bigger allocation of grass here. There are 8ha in each of these blocks while B and C are smaller at 6ha each. There are some cows that never go to C or D,” Caroline says.

Because of the time of day that they are grazed at and the length of time spent in each zone, the amount of grass being allocated in each section is different. This is where Caroline’s skill at managing the grass comes into play. Effectively, she has to juggle the allocation around to the four different parts of the farm.

Allocate too much in one area and the cows will have no incentive to leave the paddock. Allocate too little and all the cows will be in the yard too early and at the same time. At the moment, cows are getting 6kg DM grass in blocks A and D, 2.5kg DM in block B and 3kg DM in block C. They are being fed 0.5kg of meal in the robot. Post grazing heights are between 5cm and 6cm, so clean out is not as good as one would like.

To help get a better understanding of where the cows are at different times, 30 solar powered cow location transmitters are dotted around the farm. The data from these stations is used by the researchers to plot where the cows are at different times to avoid bottlenecks at the robot.

It is avoiding bottlenecks that is the key to the system, but at the same time the robot should be working for as much of the day as possible. Caroline says that at best, the robot should have about 10% idle time, that’s when it is not milking, washing or attaching clusters.

When cows were milking at peak production back in May, the robot still had 20% idle time so the researchers are planning to milk about 90 cows on the robot next year and still avoid going under 1.5 milkings per cow per day.

Daily routine

Typically, Caroline arrives on to the farm at 8am and the first thing she will do at that point is to change the filter sock and wash down the yard. After that, she will go into the office and check reports and then put up two wires, one in D and maybe B too if there are no cows left grazing there. Importantly, she never fetches cows as the cows need to make the decisions for themselves.

After these essential tasks are complete, the next few hours are her own where she can do a grass walk or catch up on maintenance or book work. In the evening, the filter sock needs to be changed again and two more wires need to be put up.

Cows are currently milking 1.52kg MS/cow/day and at peak they milked 1.78kgMS/cow/day. The difference between the peak and now is low at 0.26kg MS while on most other farms it is at least twice that. It looks as though the production curve will have a low and relatively flat peak. Cumulative milk production to 25 September has been 370kg/MS/cow. Total meal fed to date is 330kg/cow.

Labour use

A study by Teagasc on labour use on commercial farms was conducted between March 2014 and February 2015. Ten of the farms were conventially milking between 70 and 160 cows (average 120) and seven of the farms were automatically milking between 69 and 205 cows (average 105). Just under 16hrs/cow/year was spent working on the robot farms, while 25hrs/cow/year was spent working on the cows with conventional milking.

The difference in hours worked between the two systems is significant with time saved at milking being the biggest difference between the systems. However, on the robot farms more time was spent allocating grass per day.

What the robot study proves to me is that the technology and the knowledge needed to use it is available to make robotic milking work on a grass-based, spring-calving farm.

But does that mean robots are the way forward?

We know that the capital costs of a robot at about €200,000 for all components and gates is much greater than a medium-spec conventional parlour and far greater than a low-spec parlour.

We also know that a medium-spec parlour is more profitable than a robot at both 70 and 140 cows, even with all paid labour. This means that the competitiveness of robot farms will be less than that of conventional farms. With volatile milk prices farmers should be increasing competitiveness not reducing it.

Of course, it is difficulty sourcing labour that has many farmers looking at the robot option.

And while spending 16 hours work/cow/year is attractive, it must be said that there are some farmers in Ireland, albeit a minority, achieving this with conventional milking systems. The other thing is that not all farmers find it hard to find employees.

Can farmers do more to set themselves apart as being employers of choice before having to spend a lot of money on automation?

Lifestyle, the time-bound nature of milking and the physical exertion required to milk cows are all factors in farmers installing robots and they are valid reasons.

But from a purely business point of view, the large amount of capital tied up, the higher running costs with robots and the skillset needed to manage them suggest to me that at current robot prices they are some way off being the solution to solving the labour problem.

Read more

Dairy walk: Tipperary farmer presents arguments for having a robot

Moorepark present results from robotic milking trial

Fendt fields a swarm of robots