En I עב
People | Site
Hello visitor | Log in
HTML tutorial HTML tutorial HTML tutorial HTML tutorial HTML tutorial HTML tutorial
research_title.jpg research_logo.gif
About Institutes Public Relations Technology Transfer Students\International Information Center (Library) Administration
Real time online NIR milk analyzer
Real time online NIR milk analyzer for determination of milk composition during milking.

Online milk analyzerReal time milk analyzer


Z. Schmilovitch1*, G. Katz2, N. Pinski2, A. Hoffman1, H. Egozi1, E. Maltz1

1- Institute of Agricultural Engineering, Agricultural Research Organization, 2- S.A.E Afimilk, Kibbutz Afikim

Consumers (public and industry) demands, as well as growing feedstuffs prices, call for a better control over production out of which milk composition and components yield are of major concern for both, producer and consumer. This in turn drives for on-line milk composition measurements that furnish the trend of precision dairy farming.

Production, health and welfare control in a modern farm is achieved by a central system, daily monitoring the incoming dataflow from a collection of independent sensors, conducting tracking and analysis enabling performance improvement from single cow to herd. Frequent data of individual cow performance are usually required for management purposes and economical assessments of the dairy operation. Milk composition; however, is information that can be used also by the milk processing plant, when composition is directing the milk to the process that yields the best economical efficiency. Today the milk that comes out of the milking parlor is stored in one tank and its composition is considered as a matter of fact. Technically, it is possible to divert milk of different compositions to different tanks according to processing priorities.

It is common knowledge that milk composition varies along lactation, between milking operations and responds to the ration fed. Therefore, at any given time the milk composition within a group of cows fed a TMR varies, in addition to individual differences dictated by genetic capacity, also according to stage of lactation and milking time. Materializing the benefit from the information about milk composition depends on certainty that the milk composition used for decision making indeed represents the physiological status of the cow on one hand and technology that enables responding to it in real time on the other. The later is available through computer controlled self feeders and variable milking frequency provided by robot milking systems (Maltz et al. 2002), the former still depends on manual sampling and of-line laboratory analysis (Maltz et al. 2004). It is quite clear that the advancement in precision livestock farming in general and that of dairy farming in particular ((Cox 2003, 2005, 2007)) are a strong catalyst towards the development of on-line milk composition analyzers that will enable to exploit the opportunities imbedded in this information. A newly developed novel on-line in-line milk composition sensor is presented. This sensor (AFILAB) measures in real time, during the milking, the concentrations of fat, protein and lactose and indicates presence of blood and SCC. The analyzers have been installed and used in several commercial and research farms for approximately two years.

The objective of this study was to develop regression model and spectral equation to predict by measured spectra content of fat, protein and lactose in the milk line of the parlor in real time of milking. An additional parallel study was conducted to evaluate the daily, between milking and during milking milk composition fluctuations. The objectives of this study were to evaluate at what frequency and sampling method the milk composition represents the physiological status of the cow in a way that can be related to for on-line decision making through frequent sampling of milk along milking, every daily milking for several days under conventional milking parlor and robot milking conditions. It was concluded that the milk composition day to day variations indicate that a single day measurement even taken weekly, can not represent the physiological and metabolic status of the high producing dairy cow and that for a satisfactory description of both physiological and metabolic status milk sampling has to include each milking and a representative sampling of the whole milkingIn the frame of anterior study in the institute of agricultural engineering, it was showed the capability of measuring composition of raw fresh milk by application of methods of spectrometry of NIR (Schmilovitch et all, 1992, 1999, 2000). On the basis of those results a prototype sensor of which used as a base to the development of semi-commercial sensor. This development was realized in the frame of collaboration SAE Afikim Company. This sensor is on going Beta-Site testing now days Automated determination of fat-corrected milk and body weight by the analyzer and walk through scale, respectively, enable to use the NRC as well as other equations for predicting dry matter intake of individual lactating cows at any given day. This introduces, in addition to the exact economic value of the milk, better evaluation of single cow performance, analysis of ratio formulation and development of group strategies. Herd genetics may be improved by selection based on frequent milk solids determination. Early diagnosis of sub-acute rumenal acidosis and ketosis by using fat, protein and weight changes in single-cow/group/herd will enable monitoring nutrition and husbandry management to shorten response for prevention and treatment.This data, in addition to currently available automated on-line data (milk yield, milk flow rate, milk electric conductivity and daily body weight) complete the set of parameters required for improving overall herd performance and alleviates management decision making.


Maltz, E., N. Livshin, S. Devir, D. Rosenfeld (2002). Using On-line Data in Management of Milking Frequency and Concentrates Supplementation in the AMS Herd. The First North American Conference on Robotic Milking, Toronto, Ontario, Canada, March 20-22, 2002:III 33-44.

Maltz, E., N. Livshin, A. Antler, I. Brukental, A. Arieli (2004). Technologies and Modeling, for Precision Protein Feeding of Dairy Cows during Transition Time. 2004 CIGR International Conference, Olympics of Agricultural Engineering, 11-14 October, 2004, Beijing, P. R. China.

Cox, S., Editor. Precision livestock farming, 4th European Conference on Precision Agriculture (4th ECPA) and 1st European Conference on precision Livestock Farming (1st ECPLF), Berlin, Germany 15 – 19 June 2003.

Cox, S., Editor. Precision livestock farming ’05, Implementation of Precision Agriculture, 2nd ECPLF, 9-12 June 2005 in Uppsala, Sweden.

Cox, S., Editor. Precision livestock farming ’07, Proc. 3rd ECPLF, 3-6 June 2007, Skiathos, Greece

Schmilovitch, Z., E. Maltz, and M. Austerweill. (1992). Fresh raw milk composition analysis by NIR spectroscopy. in Proc. Int. Symp. on Prospects for Automatic Milking. Wageningen. The Netherlands. EAAP Publication No.65, 193-198.

Schmilovitch, Z., I. Shmulevich, A. Notea and E. Maltz. (1999). NIR sensing of a heterogeneous agricultural fluid product. NIR 99, Annual International conference of NIR. Verona Italy, June 99.

Schmilovitch, Z.,  E. Maltz, A. Hoffman, H. Egozi, Y. Belnky, I. Shmulevich and  A. Notea. (2000). Low Cost Near Infrared Sensor for On-Line Milk Composition Measurement. roceeding of The XIV Memorial CIGR World congress 2000. Tsukuba, Japan

19/05/08 | Monday
No comments (0