This presentation is based on a paper by Mein & Thompson (1993) produced for a seminar on "Managing the 30,000 pound herd" organized by the American Dairy Science Association in 1992. A number of interesting papers on the science or practice of milk harvesting have been published in the intervening 6 years. The new information supports some of the conclusions we presented in 1992 but other conclusions need to be modified now.

Compared with the challenges of feeding and breeding management for a herd of 100 lb/day cows, we (Mein & Thompson, 1993) concluded that milking was likely to be the easiest part of managing such a herd. The top cows in excellent herds were producing at or above this level six years ago, and they were being milked successfully.

Second, we concluded that the principles for milking the 100 lb/day cow were the same as for any other dairy cow: her teats should be clean and dry for milking; she should be milked gently, quickly and completely with minimal machine stripping or overmilking. We said, however, that the application of these principles might be different because high producing cows have:

1) a lower pre-milking stimulus requirement compared with low producing cows;

2) higher peak milking rates and higher average flowrates (but, in spite of this, high producers take longer to milk out);

3) a higher incidence of teat-end lesions (the most common being hyperkeratosis);

4) higher risk of new mastitis infections.

Third, we stated that existing national and international standards for construction and performance of milking systems may not be adequate to cope with the higher expected flowrates through the milking unit and milklines.

Following is a brief outline of the advances in knowledge and (my) understanding which support or challenge our 1992 recommendations and which, probably, will continue to change milk harvesting practices and systems.

In 1992, we concluded that present-day, high- producing Friesian/Holstein cows appear to need little or no manual stimulation to maximize their milk yield. Although there are no published data from cows giving 10,000 kg [22,000 lbs] or more milk per year, all the available evidence from studies with lower producing cows indicates that:

Although these conclusions are broadly correct, they now seem too simplistic. In their excellent review and analysis of research herd studies in the US and Europe, Reneau & Chastain (1995) concluded that:

Based on results of their modeling, together with research data from Rasmussen et al.(1992), Reneau & Chastain recommended that a stimulus of 10-20 sec per cow and prep-lag time of 60-90 sec results in optimum parlor throughput. It is hard to argue against a practice which reduces milking time by more than 10% and which increases milk yield by a small (1%) but significant amount. Reducing the machine-on time has important consequences for improved teat condition, as we shall see later.

In 1992, we used limited data for 16 cows in early lactation (less than 100 DIM, in 3 herds) that were producing more than 100 lb milk/day. These data demonstrated that it was not reasonable to expect high producing cows to milk out in 5 minutes flat. Although peak milk flowrates for these fresh, top cows averaged 12-12.5 lb/min (5.5-5.7 kg/min), their average milking times were relatively slow. Cows milked 2X (average milk yield = 55 lb/milking) took nearly 10 min/cow to milk while the group milked thrice daily (average milk yield = 36 lb/milking) required almost 7 min.

Other studies published since 1992 (eg., Billon, 1993; Stewart et al., 1993; Thomas et al., 1993; Reneau & Chastain, 1995) provide much better evidence for our broad guideline that, on average, cows giving 20-25 lb milk per milking should milk in about 5 min, and cows giving 30-35 lb/milking should take about 6 min. More comprehensive data are likely to come from a current project on computerized parlor monitoring lead by Dr Steve Stewart. One example given in Steve's paper for this conference shows the regression:

This equation has an R square value of 0.88. It indicates that cows in this particular herd take about 1 min longer to milk for each additional 10 lb of milk they produce per milking. Steve will be able to document, quickly and clearly, the effects of varying machine settings and milking practices on mean milking times as the project progresses.

In 1992, we used Sieber's comprehensive data (1979) to illustrate the general deterioration in teat end condition associated with increasing 305 d milk production. Cows which produced less than 5500 kg (12,000 lb) in a 305 d period had a much higher proportion of normal teat ends (30%) than higher producers. The proportion of normal teat ends fell to less than 9% for cows which produced more than 8200 kg (18,000 lb) of milk in 305 d. Sieber showed that the severity of teat end abnormalities increased and the proportion of normal teat ends decreased as the average milking time of individual cows increased from 4 min or less to 6 min or more.

Based on these highly significant changes, Mein and Thompson (1993) concluded that poorer teat end condition seemed to be an unavoidable and inevitable consequence of milking high producing cows.

I'm happy to tell you that this conclusion was premature! A Danish study (Rasmussen, 1993) has provided the springboard for a Aquantum leap@ towards better teat condition for high-producing cows milked in US dairy herds. Rasmussen=s research showed that milking time was reduced by 0.5 min per cow with no loss of milk yield when the end-of-milking threshold setting for automatic cup removers was raised from a milk flow rate of 200 to 400 gm/min. Teat condition improved markedly in the early detachment group of cows, and significantly fewer cows developed clinical mastitis.

During the past 2 years, we have been experimenting with raising the threshold flowrate from a default setting of 0.7 lb/min (for BouMatic detachers) to levels as high as 1.6 lb/min for some herds milked 3X. At the same time, the default setting of 13 sec time delay for cup removal has been shortened to 0-5 sec. The net effect has been to reduce milking times by up to 1 min or more per cow with no loss of milk yield, no change in SCC or mastitis levels. In addition to quicker milking, the major benefits have been improved teat condition and calmer cows, especially the fresh cows. The most surprising discovery of this on-going experimental work has been to find so little milk left in a typical udder following early removal. For example, average strip yield is a mere 25 ml/cow (less than 1 oz) in the herd with the highest threshold setting of 1.6 lb/min and 3 s delay time.

Results of a series of milking experiments involving high bacterial challenge showed clearly that cows which milk faster have higher infection risk (Grindal & Hillerton, 1991). These results supported earlier work in England by Dodd & Neave (1951) who noted that cows with more patent teat canals also had higher infection risk during their dry periods. Thus, high-producing cows may have higher mastitis infection risk because of the indirect relationship between high production and high milking rates (i.e. more patent teat canals).

A simulated genetic selection study in Germany indicated that incidence of mastitis will increase with increasing milk production (Jahnke et al., 1990). Given the present rate of genetic gains for milk production, new infection rates will increase by about 1% per year without positive selection for udder health in a progeny-testing program (M. Goddard, pers comm, 1998)

The likely combination of longer milk-out times and shorter pre-milking prep times in high producing herds means that milking systems should be designed with more units per operator for more efficient labor utilization. The effects of daily milk production on cow throughput are illustrated in the modelling results of Reneau & Chastain (1995). If milk yield is increased from 30 to 40 lbs per cow/milking, for example, the predicted steady state throughput falls:

• from 68 to 61 cows per hour in a D-8 parlor,
• from 80 to 71 cows per hour in a D-10 parlor
• from 91 to 82 cows per hour in a D-12 parlor
• for an optimum prep-lag time of 60 sec.

In high yielding herds, highly automated parlors should be planned to provide 20-24 units per operator, ie., a D-10 or D-12 for 1 operator, or a D-20 for 2 operators.

Economic analyses by Thomas et al. (1997) provide a convincing case for building multiple smaller parlors for 1 or 2 operators rather than a single larger parlor for more than 2 operators. Using a 15-year planning horizon, the predicted net economic advantage was almost $3 million if two D-20 parallel parlors were built instead of one D-40!

In the six years since 1992, national and international standards for the construction and performance of milking machine installations have been revised extensively. The revised standards incorporate new performance specifications to provide a common basis for evaluating the great variety of types and sizes of milking systems used throughout the world. For the first time, the main performance specifications are the same for international (ISO 5707:1996) ) and US (ASAE 518:1996) standards.

These standards should produce cost-effective improvements in milking and cleaning performance in many milking systems. The new performance criteria can be used, with confidence, to evaluate the system components of existing milking installations. For designing new installations, both the ASAE and ISO standards include guideline tables of recommended sizes for system components such as vacuum pumps, airlines and milklines.

I am happy and relieved to tell you that the number of telephone enquiries on these system design and capacity issues has fallen to a mere trickle now. Similarly, telephone enquiries on how to improve the performance of vacuum regulators have dropped off. By now, most people have figured out how to improve the dismal and expensive inefficiencies of vacuum regulation systems that were so common up to 1996. In most installations, the problem has been resolved by shutting down un-necessarily large vacuum pumps and mounting the regulator sensor closer to the sanitary trap.

Currently, the four most common issues and concerns about system components are:

What is the "right" vacuum level for milking?

It is common knowledge that increasing the system vacuum level results in faster milking times. Carefully conducted research studies have shown that milking can be successful with vacuum settings as high as 20" Hg. On the other hand, outbreaks of clinical mastitis, high cell counts, or poor teat condition frequently are linked with an unsuspected high vacuum problem on commercial dairy farms.

It seems that we must reach a compromise between machine settings for fast milking and for maintaining healthy teats and udders because the benefit of higher vacuum level may be offset by higher strip yields, higher incidence of hyperkeratosis at the teat orifice, and more machine-induced teat congestion and edema.

According to guidelines in the Milking Maching Manufacturers' Council booklet "Maximizing the Milk Harvest", the system vacuum should be set between 12.5" and 13.5" Hg (42-45 kPa) for lowline milking, and between 14" and 15" Hg (47-50 kPa) for highline milking. Usually, this will result in a mean claw vacuum within the range 10.5-12.5" Hg (35-42 kPa) during the period of peak milk flow for a representative sample of cows. Lower mean values result from problems such as excessive milkline height, restrictions in the milk hoses, excessive vacuum drop across ancillary components, blocked air vents, excessive air admission through air vents, or air leaks into the cluster.

As a general recommendation, the system vacuum should be set so that mean claw vacuum during peak milk flow is about 12-12.5" Hg (40-42 kPa) to milk cows as quickly as possible, but still within the range recommended to maintain gentle milking conditions. Nevertheless, provided that cow preparation procedures are excellent and the milking system is adjusted to minimize the low flow rate period of milking, I think we can experiment with raising the mean claw vacuum level above 12.5" Hg. This implies that parlors with low-level milklines could be set to a system vacuum above 13.5" Hg (45 kPa) and highline systems could be set above 15" Hg (50 kPa) under certain circumstances. My advice to our dealers is to optimize individual milking systems, before raising the system vacuum, according to the following steps.

When the ACR settings are optimized for a given farm, the system vacuum could be raised about 0.3-0.5" Hg (1-2 kPa) per week. The system vacuum should NOT be set above 15" Hg if:

- an Air Star pump is used, because this voids the pump warranty.

- ACR's are not installed and/or not optimized.

- milking procedures are inadequate.

The veterinary consultant should be responsible for deciding if pre-milking hygiene and management procedures are adequate to reduce the greater risks associated with milking at higher vacuum level.

A competent observer (from the dealership or the veterinary consultant) should be responsible for monitoring any changes in the mean strip yields per cow within 1 week after the system vacuum is raised, and watching for any changes in teat condition during the month after such a change.

ASAE S518:1996 Milking Machine Installations - Construction and Performance. American Society of Agricultural Engineers, St Joseph, MI.

Dodd, F. H. and F. K. Neave. 1951. Machine milking rate and mastitis. J. Dairy Res. 18: 240.

Grindal, R. J. and J. E. Hillerton. 1991. Influence of milk flow rate on new intramammary infection in dairy cows. J. Dairy Res. 58:263-268.

ISO 5707:1996. Milking machine installations - construction and performance. International Standards Organization, Geneva, Switzerland.

Jahnke, B. J., J. Wolf and D. Dietl. 1990. The inclusion of udder health in selection of cattle. pp 258-266 in Symposium, Populationsgenetische in der praktischen Tierzucht, Leipzig, Germany.

Johnson, A.P., G.A. Mein, and L. Petersson. 1996. Procedures for Evaluating Vacuum Levels and Air Flow in Milking Systems. Proc. 35th Annual Meeting, NMC, Nashville, TN, USA, p204.

Mein, G.A. and P.D.Thompson. 1993. Milking the 30,000-pound herd. J. Dairy Sci. 76:3294-3300.

Rasmussen, M.D., E.S. Frimer, D.M. Galton and L.G. Petersson. 1992. The influence of premilking teat preparation and attachment delay on milk yield and milking performance. J. Dairy Sci. 75:2131.

Rasmussen, M.D. 1993. Influence of switch level of automatic cluster removers on milking performance and udder health. J. Dairy Res. 60:287-297.

Reneau, J.K. and J.P.Chastain. 1995. Pre-milking cow prep: adapting to your system. Proceedings, National Mastitis Council Regional Meeting, Harrisburg, PA.

Sieber, R. L. 1979. Bovine teat end lesions: the prevalence and relationship to mastitis and machine milking. M. Sci thesis, University of Minnesota, USA.

Thomas, C.V., M.A. DeLorenzo and D.R. Bray. 1993. Predicting individual cow milking time for milking parlor simulation models. J. Dairy Sci. 76:2184.

Thomas, C.V., M.A. DeLorenzo, D.R. Bray, R.N. Weldon, R.A. Bucklin and J.G. Martin. 1997. A stochastic economic analysis of large herringbone and parallel parlors. J. Dairy Sci. 80:2418-2428.