¹Quality Milk Promotion Services, Department of Population Medicine and Diagnostic Sciences, Cornell University,
² Research Institute for Cattle, Sheep and Horses, Lelystad,, the Netherlands,
³Fullwood Fusion, Wijk bij Duurstede, the Netherlands


Introduction

Harvesting milk from dairy cows is the most important job on the dairy farm. The largest component of dairy farm income is generated during the milking process. Therefore, milking is a key component of managerial tasks on a dairy farm. In smaller dairies, often the owner performs most of these tasks, but on larger farmers personnel is hired to perform this function. In a lot of industries, robots have taken oven jobs that are repetitive and can be programmed by computer. The challenge in the milking process is that a live individual, the cow, needs to be milked, and that the robot needs to adapt itself to a continuous changing set of parameters. This is very unlike most robotic processes, where virtually everything can be maintained at virtually identical localization and of identical size. However, many advantages have been made in robotic milking, and robotic milking has obtained a true spot for itself in the market for milking machines of dairy cows (Anonymous 1998). There are no major concerns anymore about the question 'will it work', the answer is a very clear 'yes'. Mechanical problems are more or less solved, but the milking process is more than just harvesting a raw product. It is also a period of close contact between the milker and the cow where diagnosis on abnormalities of the cow or the milk can be made, and milking requires a set of skills that result in a product of lower or higher milk quality. In robotic milking these tasks will also be assigned to the robot, and it is specifically these issues that are of importance to the producer, the milk buyers, regulators, and ultimately the consumers.

In this presentation, we will attempt to summarize some of the literature on the topic of milk quality, and detection of abnormal milk or abnormal health status of the cow.


Uptake of Robotic Milking

Robotic milking is currently finding its way onto the dairies in Western Europe. Essentially, farmers building a new farm, or remodeling their milking parlor are considering the use of robotic milking. Currently approximately 400 systems are installed, and the industry has a hard time to keep up with demand. Countries were important number of robotic milking systems have been implemented are: the Netherlands (+200), Germany (+100), Denmark (+50), and France, Belgium and the United Kingdom (+25)(Anonymous 1998). In the Netherlands, in 1998 of all farmers building new barns, 25% put in a robotic milking system, and also 25% of all new milking systems sold into existing barns was a robotic milker. Most major milking equipment companiesnow have a robotic milker in their program, or are planning on having a version on the market very soon. The two models currently dominating the market are the 'Lely' system (also adopted in an own version by Fullwood Fusion), and the Prolion system (also adopted in an own version by Manus and Gascoigne). The Lely/Fullwood system is a single stall system, whereas the Prolion/Manus/Gascoigne system is a multi stall system. Depending on the number of stalls, the adaptations to the old facilities etc., the price per stall is between $US 100,000 to $150,000. Approximately 60 cows can be serviced by a single stall. Recently, some new barn designs were introduced that resulted in an investment per cow that was approximately similar between conventional milking parlors, and the robotic milking system. Efficient ways to guide cow flow, and optimal use of space resulted in considerable savings in the remainder of the building to make the robotic milker relatively comparable in cost to the traditional milking systems. When robotic milking leads to an increase in milk production (essentially due to 3X milking), and labor that has become available due to fewer milking hours can be used in a profitable way, robotic milking has proven to be cost-effective in smaller herd sizes in Europe (van Scheppingen and Nijssen 1998).


The uptake of the cows of robotic milking is relatively well. In most practical situations, the adaptation process takes approximately 2 months. A number of cows have continuous problems with their milking frequency, and need to be manually steered towards the robot. Also a number of cows have such udder shapes that attachment of the cluster by the robot is virtually impossible. Research has shown that in 90% of cases the attachment of the cluster was successful in the first attempt, the remaining 10% needing an additional try. In a number of commercial farms it was estimated that about 5-10% of cows needed to be replaced when the robot took over the milking procedure (Dorresteijn 1998). In some cows, the placement of the teat is such that attachment is not possible. In young animals with a lot of udder edema this may make milking impossible. Either attachment by hand, or surgical correction of teat placement is required to milk such cows. Hence, in the long term, selection pressure on cows will include the correct placements of teats to be able to have the cluster attached by robot.

Uptake of robotic milking by regulatory officials is still in its early stages. In some countries there is no problem to have cows milked by robot, in others there is a law that prescribes that all cows are checked for cleanliness and mastitis before attachment of the cluster. Strictly spoken, in the latter case, there should be a person watching over the milking during 24 hours a day (Anonymous 1998). The precise implementation of the regulations on farms with robotic milking is still not completely resolved.


Milk Quality Issues

Milk quality concerns on farms with robotic milking have been suggested with regard to clinical mastitis detection, spread of contagious subclinical mastitis, increase in plate loop count, acidity of milk, and possibly problems with freezing point of milk. In a relatively large study these issues were addressed when milk quality was compared between farms with robotic milking systems, farms with 2X milking in a conventional systems, and farms with 3X milking in a conventional system (Klungel et al. 1999). The results of this study are shown in Table 1.


Table 1. Comparison of milk quality between 28 farms with robotic milking, and conventional milking using a 2X or 3X system. Results from Klungel et al. 1999.

Parameter	Robotic Milking	Conventional 2X	Conventional 3X
Number of Farms	28	49	28
Somatic Cell Count (cells / .01 ml)	233	178	169
Plate Loop Count (cfu / .01 ml)	18.2	8.7	7.5
Freezing Point (°C)	-.517	-.520	-.523
Aciditi (mmol/100 gr fat)	0.55	0.45	0.48

Somatic cell count was not really a problem in the farms with robotic milking. Cell counts before introduction of the robot were already higher in these farms, and after introduction of robotic milking they tended to decrease. This is shown in figure 1.

In this study there was clearly no negative effect of robotic milking on somatic cell counts. There was no obvious problem with contagious mastitis in these herds, although in some situations an outbreak of contagious pathogens has been reported after introduction of robotic milking. Also clinical mastitis detection in these herds did not results in problems as judged by bulk milk somatic cell counts. Detection based on milk conducivity is used in the robot milker, and studies have shown that electrical conductivity changes are good indicators to detect clinical cases of mastitis (Nielen et al. 1992).

In a similar opposite pattern, plate loop counts increased after introduction of robotic milking. This was true for most herds, and in some herds this resulted in a number of extra penalties due to plate loop counts over legal limits. Causes for an increase in plate loop counts may be the continuous operation of the milking units, a less then optimal frequency of cleaning, long pipelines between the milking unit and the bulk tank, and the difficulty of cooling milk during and after the cleaning process of the bulk tank. Since milking takes places during the whole day, it takes a relatively long interval before the bulk tank has been filled to approximately 10% of its capacity and cooling starts. Milk going into the bulk tank immediately after cleaning of the tank will not be cooled until the tank has filled up, and this may lead to growth of micro organisms.

Figure 1. Somatic cell count development per month after introduction of robotic milking (n=28 herds). Results from Klungel et al. 1999.


On farms with robotic milking freezing point decreased towards zero. The reasons for this might be a higher amount of added water due to more frequent cleaning of the system, and immediate use of the system after cleaning so that no drying of the units occurs. Although freezing point changed, it was not a major concern on the farms where robotic milking was introduced.

Acidity of milk, indicating the amount of free fatty acids in milk increased. It is expected that frequency of milking may have an effect on acidity of milk (3X conventional milking also showed a higher acidity of milk), and also the attachment method that is used in robotic milking may have an effect on this (Ipema and Schuiling 1992). Since air inlet at attachment is relatively high in robotic milking, this may have an effect on the fatty acid composition of the milk.

Discussion

Robotic milking is certainly on its way to a larger part of the market share in milking equipment. Especially in countries with relative small herd sizes, and high labor costs, robotic milking is a viable alternative to conventional milking parlors. Most technical problems as far as attachment and operation seem to be overcome at this point in time. Current concerns are on milk quality, efficiency, and economical cost-benefit ratio.

There are relatively few studies that have compared milk quality between farms with conventional milking and robotic milking. The studies that are reported using indicate an increase in plate loop count, and no of very little effect on somatic cell count. Plate loop counts can be decreased by adapting the cleaning procedures, increasing the number of cleaning cycles per day, and by solving the bulk milk cooling problem for the initial milk after cleaning. There are a number of options for cooling of milk immediately after cleaning of the bulk tank. A small buffer tank that takes the first milk after cleaning, and that is much earlier filled to capacity to be able to start cooling is a frequently used option. Also, cooling in-line between the robotic milker and the bulk tank to decrease milk temperature before milk gets into the bulk tank is utilized, finally, different cooling systems that allow immediate cooling of small quantities of milk are available. These solutions may help in reducing plate loop counts in herds with robotic milking. Increasing the number of cleaning cycles per day is another option, but this will result in a decrease of efficiency of the robotic milking system.

Figure 2. Plate Loop Count development per month after introduction of robotic milking (n=28 herds). Results from Klungel et al. 1999.


Current research in robotic milking focuses on cleaning systems of the milking equipment and the bulk tank, and on improvement of the sensors in the robot. Since a relative small number of clusters are used to milk all cows in the herd, it is feasible to invest in sensors that will detect abnormalities in the cow and the milk. Future developments in better detection of abnormal milk, detection of heat, and analysis of milk components, or abnormal milk constituents (such as ketone bodies) may be expected in the near future. These high quality sensors will replace the intimate contact between milker and cow, and still lead to an ability to detect cow or milk deviations from normal.


References

Anonynous. Das Management ist noch wichtiger als beim Melkstand - Melkroboter. Top Agrar 1999.

Dorresteijn, J. Experiences with robotic milking: a practical approach. Report of seminar on robotic milking, April 1998, Research Institute for Cattle, Sheep and Horses, Lelystad, the Netherlands.

Ipema, A.H., and E. Schuiling. Free fatty acids, influence of milking frequency. P 491- 496 in Prospects for automatic milking. EAAP publication no. 65. 1992.

Klungel, G.H., Slaghuis, B.A. and H. Hogeveen. 1999. The effect of introduction of automatic milking systems on milk quality. Submitted.

Nielen, M., H. Deluyker, Y.H. Schukken, and A. Brand. Electrical conductivity of milk: measurement, Modifiers and Meta analysis of mastitis detection performance. J. Dairy Sci. 75:606-614.

VanScheppingen, T., and K. Nijssen. An economic look at the milk robot. . Report of seminar on robotic milking, April 1998, Research Institute for Cattle, Sheep and Horses, Lelystad, the Netherlands.

Material provided courtesy of National Mastitis Council. Used with permission. National Mastitis Council Regional Meeting Proceedings(1999) pp 64 - 69.