A quarter milking analysis device - Development and demonst…

b i o s y s t e m s e n g i n e e r i n g 1 4 7 ( 2 0 1 6 ) 2 5 9 e 2 6 4

Available online at www.sciencedirect.com ScienceDirect

journal homepage: www.elsevier.com/locate/issn/15375110

Research Paper

A quarter milking analysis device e Development and demonstration

John Upton a , * , Douglas J. Reinemann b , John F. Penry c , Paul D. Thompson b

a Animal and Grassland Research and Innovation Centre, Teagasc, Moorepark, Fermoy, Co. Cork, Ireland b Biological Systems Engineering Dept., University of Wisconsin-Madison, Madison, WI 53706, USA c Department of Dairy Science, University of Wisconsin Madison, Madison, WI 53706, USA

a r t i c l e i n f o

The objective of this paper was to describe and demonstrate a novel quarter milking analysis device (Mi4) for use in the field of bovine milking research. The Mi4 was developed for investigating the effects of milking machine factors such as vacuum level, pulsation settings and liner characteristics on milk flow rate (MFR) during the peak flow period from individual teats within a single milking. A method to use MFR, teat-end vacuum and pul- sation characteristics measured with this device was developed to estimate teat-end tissue congestion. The changes in teat-end thickness, and changes in teat canal cross sectional area, induced by congestion are small, hence a high precision device is required to detect the resulting subtle changes in MFR. No device has been described in the literature that can apply multiple experimental milking treatments within a single milking while simulta- neously recording quarter level response data. This within-milking approach accounts for non-linearity as total milk mass increases and is independent of any effect of milk hose slope and positioning. Sources of variability exist in MFR within an individual quarter, from quarter to quarter, between cows and from day to day. Understanding and controlling for these natural sources of variation is critical for the successful deployment of the Mi4 in milking experiments. Experimental results demonstrate that the Mi4 equipment and methods can detect changes in quarter MFR as small as 2% by applying treatments within the peak flow period of individual quarters, thus making it a useful tool for milking ma- chine research. © 2016 IAgrE. Published by Elsevier Ltd. All rights reserved.

Article history: Received 8 November 2015 Received in revised form 7 April 2016 Accepted 29 April 2016

Keywords: Milk harvesting Quarter milking Milking management

response of the animal to milking ( Gleeson, O ' Callaghan, & Rath, 2003; Marnet & McKusick, 2001; Reinemann, Bade, Zucali, Spanu, & Ruegg, 2008 ) and to indicate the efficiency of milk ejection ( Tancin & Bruckmaier, 2001 ). Moreover, un- derstanding how milking machine settings influence the



Milk Flow Rate (MFR) and yield recording is typically done at the udder level rather than the quarter level. Analysis of MFR at the udder level has been used to study the physiological

* Corresponding author . E-mail address: john.upton@teagasc.ie (J. Upton).

http://dx.doi.org/10.1016/j.biosystemseng.2016.04.016 1537-5110/ © 2016 IAgrE. Published by Elsevier Ltd. All rights reserved.


b i o s y s t e m s e n g i n e e r i n g 1 4 7 ( 2 0 1 6 ) 2 5 9 e 2 6 4

This paper describes the development and demonstration of a quarter milking analyses device (Mi4) that can be used to assess machine-induced changes in MFR using unmodified teatcups, a further novel aspect of the device. Previous work has described a method of estimating teat-end congestion using the Bernoulli equation to estimate the cross sectional area of the teat canal ( Reinemann et al., 2008 ). The change post-milking compared to pre-milking in teat-end thickness induced by congestion has been shown by ultrasonography to be about 5% different ( Spanu et al., 2008 ). The changes in the cross sectional areas of the teat canal were hypothesised to be of similar magnitude. The equation used to estimate changes in teat canal area requires measurement of MFR (kg min  1 ), the milk ratio (MR) or fraction of the pulsation cycle during which milk is flowing, and the pressure difference across the canal. The cross-sectional area of the teat canal (CA, m 2 ) is estimated using the following equation ( Upton, Penry, Rasmussen, Thompson, & Reinemann, 2016 ): CA ¼ a  Q  MR  1  ð Vsmtb þ 4500 Þ  1 = 2 (1) where a is a constant (28.4 (kg m  3 ) 1/2 ) incorporating the fric- tion factor of the teat canal and the length:diameter ratio of the canal and Vsmtb (Pa) is the average short milk tube (SMT) vacuum during the milking phase of the pulsation cycle. A positive pressure in the udder cistern of 4500 Pa was assumed ( Bruckmaier & Blum, 1996 ). Q is the volumetric flow of milk (m 3 s  1 ) where r is the density of milk (1030 kg m  3 ). This paper describes two experiments to quantify the sources of variance for each of these terms and methods used to reduce these sources of variance to obtain an estimate of MFR and hence, teat canal cross-sectional area, to within ± 2%. In addition to estimating teat-end congestion the device was also designed to measure other aspects of milking perfor- mance including Vmpc and air admission to teatcups at the quarter level. Q ¼  MFR 60  r (2)


treatment milk flow rate (kg min  1 )


CA canal area (m 2 ) MFR milk flow rate (kg min  1 ) MPC mouth piece chamber MR milk ratio n sample size O control milk flow rate (kg min  1 ) OP

over pressure of the milking liner (kPa)

udder MFR is important for the development of best practice parameters for machine settings and for appropriate sizing of milking facilities ( Jago, Burke, & Williamson, 2010; O ' Brien, Jago, Edwards, Lopez-Villalobos, & McCoy, 2012; Rasmussen, 1993 ). There have been relatively few studies that have measured MFR at the quarter level. One such study reported the use of an ultrasonic flow meter which was fitted in the short milk tube of a conventional teat cup ( Delwiche, Scott, & Drost, 1982 ). The authors claimed to measure instantaneous changes in milk flow-rates from a single teat during the course of machine milking. Use of an optical flow meter in the short milk tube of a modified teat cup was reported by Williams, Mein, and Brown (1981) . Both of these devices required considerable modification to a conventional teatcup. Experiments have also been carried out using artificial udders with flow rates controlled at the quarter level ( O ' Callaghan, 2002, 2004 ). This approach can be used to explore the effects of milk flow rate on the vacuum dynamics of the cluster, but cannot provide information about machine- induced changes on MFR. A variety of devices also exist for quarter level milking-time tests of mouth piece chamber vacuum (Vmpc), pulsation chamber vacuum (Vpc) and short milk tube vacuum (Vsmt) (e.g. Vadia, Biocontrol AS, Norway) but lack the ability to simultaneously record MFR. Quarter level studies have been done correlating quarter MFR with somatic cell count ( Tan  cin, Ipema, & Hogewerf, 2007 ), stage of lactation, cow parity, time of milking and quarter position ( Tan  cin, Ipema, Hogewerf, & Ma  cuhov  a, 2006 ) and milk composition ( Forsb € ack et al., 2010 ). Recently, greater emphasis has been placed on quarter milk flow patterns because they offer biological information that is needed for improving machine milking and the welfare of cows ( Tan  cin et al., 2006 ). PC pulsation chamber PMF peak milk flow Q volumetric flow of milk (m 3 s  1 ) SMT short milk tube Vmpc mouth piece chamber vacuum (kPa) Vs system vacuum (kPa) Vpc pulsation chamber vacuum (kPa) Vsmt short milk tube vacuum (kPa) Vsmtb average short milk tube vacuum during the milking phase of the pulsation cycle (Pa) a constant of 28.4 (kg m  3 ) 1/2 r density (kg m  3 ) s 2 sample variance (kg min  1 ) 2


Materials and methods


Device development

The Mi4 consists of four teat-cup and liner assemblies, a vacuum regulator (Model 487, L.J. Engineering Inc., CA, USA) and a pulsation system (pulsator type V658170-1 De Laval, Tumba, Sweden). A schematic of the sensor placement on the device is shown in Fig. 1 . For the purposes of this test a Milkrite IP4-LM liner and shell combination was used (Milkrite, Wilt- shire, UK), although any liner and shell combination could be used. The milk from each quarter is collected in a cylindrical tube, suspended from an s-beam load cell (Omega Engineering Inc., CT, USA). Miniature vacuum transducers (40PC series, mini signal conditioned vacuum sensors, Honeywell, NJ, USA) were fitted into the mouth piece chamber (MPC), SMT and pulsation chamber (PC) of each liner. National Instruments


b i o s y s t e m s e n g i n e e r i n g 1 4 7 ( 2 0 1 6 ) 2 5 9 e 2 6 4

total pre-milking preparation process took between 60 and 90 s prior to teatcup attachment.

2.2.2. Experiment 2: effect of pulsation settings on milk flow rate Six cows were milked with the Mi4 while alternating pulsation settings during a single milking. The pulsator-off time was alternated between 250 ms and 375 ms (pulsator ratios of 0.70 and 0.60 respectively) for eight complete pulsation cycles during the peak milk flow (PMF) period. All other settings and teat preparation were as described in 2.2.1. This change in pulsation settings was chosen to create a change in MFR of about 15% to test the ability of the Mi4 equipment and methods to detect changes in MFR. The criterion for calculating MR was that Vpc was greater than OP producing an estimated resolution of the average MR for 5 pulsation cycles of about 0.1%. The resolution of the average Vsmt during the milk phase (about 2500 measurements for each 5 pulsation cycle interval) was also estimated to be 0.1%. The limiting factor in achieving the ability to detect < 2% change in CA is the accuracy of the MFR measurement. MFR was calculated for the last 5 full pulsation cycles (defined as an interval) of each 8 pulsation cycle treatment period. This strategy was adopted to allow vacuum levels and MFR to stabilise during the first 3 pulsation cycles after changing treatment conditions. Data were compiled and analysed using SAS 9.4 (SAS Institute Inc., NC, USA). The PMF period was defined as when MFR was within 80% of the maximum five pulsation cycle interval MFR. Portions of the flow rate curve before or after the PMF period were removed from the analysis. The average duration of the PMF period was 104 s. Quarters without a PMF period of at least 60 s were removed from the analysis. The minimum, mean, maximum and variance of MFR by cow and quarter during the PMF period were calculated. The ‘Proc Nested ’ statement of SAS was used quantify the contribution of day, cow, quarter and within- quarter interval to the overall variance in MFR in the dataset. 2.3.1. Treatment effect on milk flow rate The analysis carried out in experiment 2 followed the same data processing and screening procedure as described in Section 2.3 . The GLM (General Linear Model) statement of SAS was used to test for significant differences between mean quarter MFR values of the control treatment and the longer pulsation off-time treatment. The model used was MFR ¼ treatment, with treatment declared as a class variable. 2.3. Data analysis and screening

hardware (NI Compact Rio 9024 chassis, with NI9237, NI9205, NI9263 and NI9481 C-series modules) and software (Labview 2014) (National Instruments, TX, USA) were used for control- ling pulsation timing and vacuum level and for recording Vmpc, Vsmt, Vpc and system vacuum (Vs). All data were recorded with a sample rate of 1000 Hz. The milk mass measurement system was calibrated using a four-point calibration routine. MFR was calculated by linear regression of milk mass over time for 5 complete pulsation cycles as MFR within a pulsation cycle is pulsatile. MR was calculated as the fraction of each pulsation cycle during which Vpc exceeded liner overpressure (OP) determined according to published methods ( Leonardi, Penry, Tangorra, Thompson, & Reinemann, 2015 ). Vacuum sensors were calibrated using a four-point calibration routine with reference to a calibrated vacuum recording device (Interpuls S.p.A. PT V, Italy). The accuracy of this device was verified against a mercury column. 2.2.1. Experiment 1: estimate of milk flow rate variance across cows, across days, within udder, and within quarter The Mi4 was used to milk 18 cows (72 quarters) on one day. Six of these cows were milked on two consecutive days. All of these milkings were performed under constant milking con- ditions with a Vs of 43 kPa, pulsation on-time of 575 ms and a pulsation off-time of 250 ms. All teats were stripped, cleaned, pre-dipped and wiped before application of the teat-cups. The Fig. 1 e Diagram of one teat-cup arrangement of the Mi4 where A ¼ pulsator, B ¼ vacuum regulator, C ¼ weigh cell, D ¼ milk collection tube, Vmpc ¼ mouthpiece chamber vacuum measurement point, Vpc ¼ pulsation chamber vacuum measurement point, Vsmt ¼ short milk tube vacuum measurement point, Vs ¼ system vacuum measurement point. Dimensions in meters. 2.2. Milking conditions



3.1. Experiment 1: milk flow rate variance across cow quarters, across udders, across days, within udder and within quarter 3.1.1. Variance analysis The mean MFR across all cow quarters was 1.09 kg min  1 (range 0.50 e 1.62). The variance in mean MFR across all cow


b i o s y s t e m s e n g i n e e r i n g 1 4 7 ( 2 0 1 6 ) 2 5 9 e 2 6 4

quarters accounted for 58.7% of the total variance in the MFR dataset ( Table 1 ). The mean udder total MFR for the cows with at least 3 quarters with PMF periods of at least 60 s was 4.14 kg min  1 (range 2.91 e 5.28). The variance in mean udder PMF across udders accounted for 34.5% of the total variance in the MFR dataset. Next by order of magnitude was the variance in within-quarter MFR, at the interval level of a quarter, which accounted for 4.80% of the total variance. Finally the variance in MFR from day 1 to day 2 accounted for 2.08% of the total variance. Table 2 shows MFR data at the quarter level. During the period when an extended pulsation off-time was applied, a mean decrease of 12% in MFR was observed. The mean control MFR across all quarters was 1.07 kg min  1 with a variance (across quarters) of 0.01 (kg min  1 ) 2 . The mean treatment MFR was 0.942 kg min  1 with a variance (across quarters) of 0.016 (kg min  1 ) 2 . The mean control MFR was significantly different from the treatment MFR (P < 0.05). The data presented showed that there are both practical and scientific advantages for employing the Mi4 device when implementing milking management research experiments. While the small sample size and limited time horizon of experiment 1 hinders the generalisability of the results, these data do demonstrate that the natural biological variance in MFR within a quarter is an order of magnitude lower than the across-cow and across-quarter variance in MFR. Applying experimental treatments within the PMF period of a single milking is a novel experimental method that controls for across cow and across quarter variability. Furthermore, ran- domising the application of experimental treatments by in- terval during the PMF period further controls for variations in MFR during the PMF period of individual quarters. For a desired detectible difference of 2% in MFR, and a within quarter variance of 0.003 (kg min  1 ) 2 ( Table 1 ), then a sample size of 39 would be required. These calculations were made using a sample size calculation for a power of 80% and a significance level of 5% ( Snedecor & Cochran, 1989 ); n ¼ 7 : 9 s 2 ð O  A Þ 2 (3) 3.2. Experiment 2: treatment effect on milk flow rate 4. Discussion 4.1. Applicability of the Mi4 to milking research

Table 2 e Milk flow rate (MFR) ( > 80% of maximum MFR of each quarter) and variance (kg min ¡ 1 ) 2 of the control treatment (i.e. vacuum of 43 kPa, pulsation on-time of 575 ms and pulsation off-time of 250 ms) and treatment MFR (i.e. vacuum of 43 kPa, pulsation on-time of 575 ms and pulsation off-time of 375 ms) within each cow quarter (CowQ) for 11 quarters. The mean control MFR (1.07 kg min ¡ 1 ) was significantly different from the mean treatment MFR (0.942 kg min ¡ 1 ), (P < 0.05). The variance of the control MFR figures across all quarters was 0.010 (kg min ¡ 1 ) 2 . The variance of the treatment MFR figures across all quarters was 0.016 (kg min ¡ 1 ) 2 . CowQ a Control MFR Variance control Treatment MFR % deviation







0.023 0.003 0.012 0.004 0.003 0.003 0.007 0.011 0.008 0.002 0.006

0.921 0.980 0.975 0.980

 4%  9%  9%  7%  9%  9%

1.08 1.08 1.05 1.24 1.25 1.08 1.06

1.13 1.13

0.924 0.962 0.749 0.809 0.791

 14%  9%  20%  18%

0.937 0.983

1.03  23% a 1FL ¼ cow 1, front left, FR ¼ front right, RR ¼ rear right, RL ¼ rear left.

where n ¼ sample size, s 2 ¼ sample variance, O ¼ control MFR and A ¼ treatment MFR. Clearly, the decision criteria around identification of the PMF period influences the number of quarters that proceed to the final statistical analysis. These criteria were set with knowledge of the stable peak milk flow profile realised during bovine milking provided cows are stimulated before cluster attachment ( Ambord & Bruckmaier, 2009; Kaskous & Bruckmaier, 2011 ). Hence, our description of the PMF period as being within 80% of the maximumflow-rate may even have been too coarse. Tightening this tolerance may increase the difficulty in identifying a suitable number of experimental quarters, but would result in a lower population variance. Quarters intended for use in milking management experi- ments with the Mi4 should be analysed prior to treatment application (i.e. as described in experiment 1) for levels of inherent MFR variance. Hence the sample size required for statistical significance may be refined iteratively prior to experiment implementation, which would ensure robust results. The Mi4 device gives researchers the ability to advance the state-of-the-art in milking management knowledge through high accuracy data acquisition and precise application of milking management conditions during the PMF period. While the results presented here describe detailed flow-rate data, there is much potential knowledge to be gained by coupling flow-rate data with vacuum recordings in the PC, MPC and SMT at the quarter level. One pertinent application of the data could be in understanding the dynamics of teat-end and teat- barrel tissue changes during milking and how these dynamics influence MFR. For example teat thickness changes have been

Table 1 e Sources of variance within the milk flow rate database generated in this experiment. Variance source Variance (kg min  1 ) 2 % of total variance

Across quarter

0.033 0.020 0.003 0.016

58.7% 34.5%

Across cow

Within quarter

4.78% 2.08%

Across day


b i o s y s t e m s e n g i n e e r i n g 1 4 7 ( 2 0 1 6 ) 2 5 9 e 2 6 4

documented in response to specific milking machine setting changes ( Gleeson, O ' Callaghan, & Rath, 2004; Hamann & Mein, 1996 ). Changes in teat dimensions are observed after milking because the forces generated by the milking vacuum, which cause the teat to stretch both radially and longitudinally, induce congestion and oedema in the teat wall ( Mein & Reinemann, 2014 ). However it would be interesting to quan- tify the effect of these changes (if any) on MFR values and hence milking times.

variation in milk yield and milk composition at the udder- quarter level. Journal of Dairy Science, 93 (8), 3569 e 3577 . Gleeson, D. E., O ' Callaghan, E. J., & Rath, M. (2003). Effect of vacuum level on bovine teat-tissue and milking characteristics. Irish Journal of Agricultural and Food Research, 42 (2), 205 e 211 . Gleeson, D. E., O ' Callaghan, E. J., & Rath, M. V. (2004). Effect of liner design, pulsator setting, and vacuum level on bovine teat tissue changes and milking characteristics as measured by ultrasonography. Irish Veterinary Journal, 57 (5), 289 e 296 . Hamann, J., & Mein, G. A. (1996). Teat thickness changes may provide biological test for effective pulsation. Journal of Dairy Research, 63 (2), 179 e 189 . Jago, J. G., Burke, J. L., & Williamson, J. H. (2010). Effect of automatic cluster remover settings on production, udder health, and milking duration. Journal of Dairy Science, 93 (6), 2541 e 2549 . Kaskous, S., & Bruckmaier, R. M. (2011). Best combination of pre- stimulation and latency period duration before cluster attachment for efficient oxytocin release and milk ejection in cows with low to high udder-filling levels. Journal of Dairy Research, 78 (1), 97 e 104 . Leonardi, S., Penry, J. F., Tangorra, F. M., Thompson, P. D., & Reinemann, D. J. (2015). Methods of estimating liner compression. Journal of Dairy Science, 98 (10), 6905 e 6912 . Marnet, P. G., & McKusick, B. C. (2001). Regulation of milk ejection and milkability in small ruminants. Livestock Production Science, 70 (1 e 2), 125 e 133 . Mein, G. A., & Reinemann, D. J. (2014). Machine milking: Volume 1- action of the teatcup and responses of the teat . CreateSpace Independent Publishing Platform . O ' Brien, B., Jago, J., Edwards, J. P., Lopez-Villalobos, N., & McCoy, F. (2012). Milking parlour size, pre-milking routine and stage of lactation affect efficiency of milking in single- operator herringbone parlours. Journal of Dairy Research, 79 (2), 216 e 223 . O ' Callaghan, E. J. (2002). Measurement of vacuum stability in milking units during simulated milking. Irish Journal of Agricultural and Food Research, 41 (2), 171 e 179 . O ' Callaghan, E. J. (2004). Effects of the design of a milking unit on vacuum variations during simulated milking. Irish Journal of Agricultural and Food Research, 43 (2), 237 e 245 . Rasmussen, M. D. (1993). Influence of switch-level of automatic cluster removers on milking performance and udder health. Journal of Dairy Research, 60 (3), 287 e 297 . Reinemann, D. J., Bade, R., Zucali, M., Spanu, C., & Ruegg, P. L. (2008). Understanding the influence of machinemilking on teat defence mechanisms. In Proc. International Conference on Mastitis Control from Science to Practice (pp. 323 e 333). The Hague, NL: Wageningen Academic Publishers NL. https://books.google.ie/books? id ¼ a39A9I3yNO0C & pg ¼ PA16 & lpg ¼ PA16 & dq ¼ Understanding þ the þ influence þ of þ machine þ milking þ on þ teat þ defence þ mechanisms & source ¼ bl & ots ¼ XPF7Ywf-Bm & sig ¼ ZqxDsml6T7nne9fI-RrVpdCukBM & hl ¼ en & sa ¼ X & ved ¼ 0ahUKEwiN5tD6mtTMAhVIBMAKHVCMDhkQ6AEIJzAB#v ¼ onepage & q ¼ Understanding%20the%20influence%20of% 20machine%20milking%20on%20teat%20defence% 20mechanisms & f ¼ false . Snedecor, G., & Cochran, W. (1989). Statistical methods (8th ed.). Ames: Iowa State Press, USA . Spanu, C., Reinemann, D. J., Momont, H., Cook, N., Ruegg, P. L., & Bade, R. D. (2008). Ultrasonic assessment of teat tissue congestion. In Proc. ASABE Annual International Meeting, Providence, Rhode Island (pp. 1893 e 1899). St. Joseph, MI: Am. Soc. Agric. Biol. Eng. Tancin, V., & Bruckmaier, R. M. (2001). Factors affecting milk ejection and removal during milking and suckling of dairy cows. Veterin  arnı´ Medicı´na, 46 (4), 108 e 118 .



A quarter milking analysis device, the Mi4, was developed and its applicability to the field of milking research was demon- strated. The Mi4 is a high precision device that can carry out milking management experiments in a novel way, through application of multiple experimental treatments to a quarter within a single milking. The usefulness of the device was demonstrated through the application of a longer pulsation off-time treatment during the PMF period which resulted in a significant reduction in MFR of 12% with a detectable differ- ence as low as 2% with a sample of 39 quarters. The ability to detect these MFR differences makes the device useful for application in the field of milking research. Finally, it was noted that quarters intended for use in milking experiments with the Mi4 should be analysed prior to treatment applica- tion for levels of inherent MFR variance; doing so would allow the sample size required for statistical significance to be computed prior to experiment implementation.


Financial support of the University of Wisconsin-Madison, Teagasc, Avon Dairy Solutions and Dairy Australia are grate- fully acknowledged. We acknowledge the assistance of the staff in the Dairy Cattle Centre at the University of Wisconsin- Madison. We acknowledge Peter Crump at the Statistical Consulting Group in the College of Agricultural and Life Sci- ences (CALS), University of Wisconsin-Madison, for statistical advice.

r e f e r e n c e s

Ambord, S., & Bruckmaier, R. M. (2009). Milk flow-controlled changes of pulsation ratio and pulsation rate affect milking characteristics in dairy cows. Journal of Dairy Research, 76 (3), 272 e 277 . Bruckmaier, R. M., & Blum, J. W. (1996). Simultaneous recording of oxytocin release, milk ejection and milk flow during milking of dairy cows with and without prestimulation. Journal of Dairy Research, 63 (2), 201 e 208 . Delwiche, M. J., Scott, N. R., & Drost, C. J. (1982). Instantaneous milk flow rate patterns from conventional teat cups. In American society of agricultural engineers summer meeting. Paper 82-3026 (pp. 214 e 218). St. Josephs, Michigan 49085: ASAE, 27 . Forsb € ack, L., Lindmark-Ma˚nsson, H., Andr  en, A.,  Akerstedt, M., Andr  ee, L., & Svennersten-Sjaunja, K. (2010). Day-to-day


b i o s y s t e m s e n g i n e e r i n g 1 4 7 ( 2 0 1 6 ) 2 5 9 e 2 6 4

Tan  cin, V., Ipema, A. H., & Hogewerf, P. (2007). Interaction of somatic cell count and quarter milk flow patterns. Journal of Dairy Science, 90 (5), 2223 e 2228 . Tan  cin, V., Ipema, B., Hogewerf, P., & Ma  cuhov  a, J. (2006). Sources of variation in milk flow characteristics at udder and quarter levels. Journal of Dairy Science, 89 (3), 978 e 988 . Upton, J., Penry, J. F., Rasmussen, M. D., Thompson, P. D., & Reinemann, D. J. (2016). Effect of pulsation rest phase duration

on teat end congestion. Journal of Dairy Science, 99 (5), 3958 e 3965 . Williams, D. M., Mein, G. A., & Brown, M. R. (1981). Biological responses of the bovine teat to milking - information from measurements of milk flow-rate within single pulsation cycles. Journal of Dairy Research, 48 (1), 7 e 21 .

Page 1 Page 2 Page 3 Page 4 Page 5 Page 6

Made with FlippingBook Annual report