Mil

H
H

109
374;
THS

 

  
   
   

  

, LIBRARY

Michigan Stab:
University

This is to certify that the

thesis entitled

The Construction of an Improved
Milk Flow Meter

presented by

Preston Miller Givens, Jr.

has been accepted towards fulfillment
of the requirements for

MS degree in Agricultural
Engineering Technology

agofiw

Major professor

 

 

Date 2/21/79

 

0-7 639

 

OVERDUE FINES ARE 25¢ PER DAY
PER ITEM

Return to book drop to remove
this checkout from your record.

 

 

 

 

 

THE CONSTRUCTION OF AN IMPROVED

MILK FLOW METER

BY

Preston Miller Givens, Jr.

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Agricultural Engineering

1979

ABSTRACT

THE CONSTRUCTION OF AN IMPROVED
MILK FLOW METER

BY

Preston Miller Givens, Jr.

A flow meter was constructed to measure the milk
production of an individual cow. This unit was designed
to be inserted in the milk hose between the milking unit
and the pipeline. The flow meter used a floating needle
valve to determine flow rate. The float was displaced
over a distance prOportional to the milk flow rate. A
modified LVDT circuit was used to sense the position of
the float and give a corresponding electronic indication
of the flow rate. Digital circuits were used to totalize
the cow's production.

Experimental results from the unit were not
encouraging. The flow meter performed reasonably well

in the laboratory but failed to give usable results in

Approved
Maior Professor

\“ )
Approved £9.r\‘ AitefléLW~4fl~

Department Chairman

the milking parlor.

 

To "Ginny Lou"

ii

ACKNOWLEDGMENTS

I would like to take this opportunity to thank
all of those people who have given me support and counsel
during my trip through the halls of academia. Special
appreciation goes to Dr. Bill Bickert and to my wife,
Jane, as these individuals have patiently endured the
frequent delays and setbacks that have characterized my

program.

iii

TABLE

LIST OF TABLES . . . .

LIST OF FIGURES . . .

INTRODUCTION . . . .
OBJECTIVES . . . . .
LITERATURE REVIEW . . .
DESCRIPTION OF OPERATION
CALIBRATION AND TESTING .
ANALYSIS AND DISCUSSION OF
CONCLUSIONS . . . . .
RECOMMENDATIONS FOR FUTURE

APPENDICES O O C O 0

OF CONTENTS

RESULTS

RESEARCH

A. SCHEMATIC DIAGRAM OF ELECTRONICS

B. CALIBRATION DATA
C. TESTING DATA . .

REFERENCES . . . . .

iv

Page

vi

10
27
32
36
37
38
39
41
43

46

LIST OF TABLES

Table Page
1. Calibration Data . . . . . . . . . 42
2. Laboratory Testing Data . . . . . . . 44
3. Parlor Testing Data . . . . . . . . 45

LIST OF FIGURES

Figure Page

 

1. Overall Mechanical Layout . . . . . . 20
2. Float Detail . . . . . . . . . . 22

3. Block Diagram of Electronics . . . . . 24

 

4. Calibration Curve . . . . . . . . . 29
5. Flow Meter Testing Data . . . . . . . 34

6. Schematic Diagram of Electronics . . . . 40

vi

INTRODUCTION

The history of mechanized milking has always been
one of incremental improvements. Generally, no one indi—
vidual has been responsible for more than a few of these
improvements. Instead, those peOple involved with the
dairy industry have contributed a concept here, a device
there, until today it is possible to equip a milking
parlor so that the operator only needs to wipe the cow's
udder, inspect it, attach the milking machine, and dis-
infect the teats after milking. Automatic machinery does
the rest: it opens and closes the parlor gates, Sprays
the udders with warm water, and removes the milking
machine from each cow when her milk flow ceases. A
parlor with this degree of mechanization was called a
”semi-automatic milking parlor" by Bickert 33 El (1972).

There is one process, however, that is difficult
to reliably automate at the present: weighing and
recording the amount of milk that each cow produces.
Under the present system Of mechanized milking, each
cow's milk production can be measured with either an
accumulating weigh jar or an in-line milk-measuring meter.

None of the devices that are available in either of these

categories are capable Of being completely automated
and giving reliable results.

An improved instrument is needed that will auto-
matically and reliably determine the amount of milk a
cow has produced and present this information in a form
that is compatible with electronic recording equipment.
The construction of such an instrument was the goal of

this research.

OBJECTIVES

The overall objective of this research was to
develop an instrument that would determine the individual
milk output of a cow at each milking.

Criteria for the instrument were:

1. To meet DHIA standards for milk weight

determination

2. To be compatible with the operation of

currently-used milking systems

3. TO be adaptable to clean-in-place operations

4. To be compatible with electronic data systems.

LITERATURE REVIEW

Man's use of milk goes far back into our recorded
history. References can be found in the Bible and in
Egyptian hieroglyphics concerning the use of Sheep and
goats' milk. Man has also been keeping account Of the
amount Of milk these animals produced and with the pro-
cess of measurement in general. Leviticus 19:36 says:
"just balances, just weights, . . . shall ye have." We
can infer from this reference that the first method of
measuring the milk production of an animal was probably
a balance beam with established weights balancing a
bucket or skin of milk. Irvin (1977) says that the
balance beam was used as far back as 3000 B.C. in Egypt.

The balance beam continued to be the common
method for weighing milk production until about A.D. 1700
when the spring scale was invented. Although the balance
beam was theoretically capable of greater accuracy, the
spring scale was easier and faster for the farmer to use
than the balance beam. The less-accurate spring scale
usually provided greater accuracy than the farmer needed.

The advent of machine milking initially brought
few changes to this picture. Most of the first machines
collected milk in a metal container over which a vacuum

4

was drawn during milking. It was a simple matter to
weigh the container both before and after the cow was
milked. The introduction of the milk-conveying pipe-
line, however, complicated this procedure. Because the
pipeline carried the milk directly to the bulk tank, it
eliminated the metal container for accumulating the milk
from each cow. Thus, there was no way to measure each
individual cow's production.

This inherent limitation of pipeline systems led
to the development of two basic types of accessories for
pipeline systems to measure the milk production of each
cow. The first type was called a weigh jar. Concep-
tually, this was the metal container of pre—pipeline
systems in that it accumulated all of the milk that a
cow produced during a milking in a large glass jar. A
new twist, however, was that the scale on the side of
the jar was calibrated in pounds rather than a volumetric
unit. This feature enabled the immediate readout of the
cow's milk production by the milker. A system of valves
on the weigh jar was used to control the flow of milk
and air into and out Of the jar, enabling the milker to
collect the milk from a cow, measure it, and then dump
it into the pipeline and bulk tank with the rest of the
milk.

The second type Of accessory developed to measure

milk production was the milk-measuring meter. These

units were essentially continuous-flow devices that
measured the flow Of milk directly as it came from the
cow without a significant amount of accumulation. They
were either a count-incremental unit, such as a Te Sa
Milk-O-Meter,* which counted the number of times a known
volume was filled with milk, or a proportional-flow
device, which divided the flow of milk by a large fixed
ratio and accumulated the smaller portion in a cali-
brated container, as in the Babson Bros. Tru-Test Meter.
Both of these types of units had their drawbacks.
The count-incremental units were well-suited to remove
and/or digital readout techniques. It was only neces-
sary to count the movements Of the mechanical linkages
inside the unit. These same mechanisms inside the unit,
however, gave rise to reliability and cleaning problems.
The damp, corrosive environment of the milking parlor
has always created trouble with moving parts, seals, and
electrical contacts. There were also some problems with
flooding Of the Milk-O-Meter under heavy milk-flow condi-
tions. Nonetheless, these problems did not prevent three
manufacturers from marketing count-incremental-type units
with remote readout capabilities. One of these units,

the Bou-Matic Milk Master, was part of a complete

 

*Mention of commercial product names is intended
only for purposes of illustration and clarification and
does not constitute endorsement by the author or Michigan
State University.

milk-production recording system that recently came onto
the market.

The proportional-flow devices, on the other hand,
had very few moving parts, mostly consisting Of one or
two valves that controlled milk flow. The internal
structure of these units made them easier to clean than
the count-incremental type, but the high temperature Of
the cleaning solutions sometimes caused distortion.
Another drawback was the difficulty in adapting meters
which used this principle of operation to remote and/or
digital readout- There was no reliable method on the
market of doing this. Even so, proportional-flow meters
have been popular with many farmers.

The primary goal of both types of milk-measuring
meters described above was the accurate measurement of
a cow's milk production. The need for this accurate
measurement stems, for the most part, from the existence
of the Dairy Herd Improvement Association (DHIA). This
group is responsible for fostering the improvement of
dairy herds through the maintenance of production and
breeding records on all cows in a herd involved in the
program. To insure some degree of control over the pro-
duction data going into these records, DHIA established
standards governing the accuracy and repeatability of
all devices that were to be used to measure milk produc-

tion. These standards were found in a document by

DHIA (1974) containing the regulations and specifications
for all facets of the record-keeping process.

The above standards established by DHIA for
milk-production measurement break down into specifications
for random errors in milk weight, bias in sampling, and
average accuracy. The random error standard says that
the errors in individual weighings must be normally dis-
tributed and that 95% of them must be within 14.1% of
the true mean. The percentage magnitude of the allow-
able error is controlled to some extent by the specifi-
cations for the amount of bias which is consistently
overweighing on some cows and underweighing on others.
The :4.1% maximum allowable error is for a bias of 25%,
but the error can be as high as :l0% when the bias is
0%. The maximum allowable bias is 25% and it is defined
as the square of the correlation between errors, expressed
as percentages of the true weight, in two milkings of the
same cow (DHIA, 1974). DHIA also specifies that 95% of
the test-group averages must differ from the true mean
by no more than 1.4% where a test-group average is the
mean performance of a single device at one milking on a
single milking system. When the above specifications
for random error, bias, and accuracy are considered
together, they clearly define the performance level
required from all DHIA-approved milk-production measur-

ing meters.

During the course of this research, one meter has
become available that meets all of the criteria for the
approval of DHIA; this approval is pending. One would
suppose that competing manufacturers are engaged in
research on similar meters, but accessibility to such
research tends to be very limited. This, combined with
the fact that the author knows of no one else in the
public sector engaged in similar research, makes ascer-

taining the current state of the art difficult.

DESCRIPTION OF OPERATION

In describing the Operation of the improved milk
flow meter, an examination of the problems involved is
in order.

The milk leaving a milking claw generally flows
in an irregular fashion that can be characterized as a
pulsating two-phase flow. The pulsations arise from
the fact that the milk is propelled through the hoses by
vacuum, but only when enough milk collects in a low spot
in the milk hose to fill the cross-sectional area of the
hose completely, thereby creating a "slug" for the vacuum
pumps to work against. This slug is accelerated rapidly
until it reaches an area where the pressure differential
which has been created across it can be released. This
can be in a pipeline, a weigh jar, or a milk meter. As
a result, a given point in a milk hose can be full, par-
tially full, or empty at any given time. This pulsating
two-phase flow makes it difficult to apply any of the
industrial flowmeters on the market, as almost all of
them require a single-phase flow for proper Operation.

In addition to the problem Of the two-phase flow,
the physical properties of milk make it quite susceptible
to foaming. Foaming will generally ocCur anywhere that

10

11

fluid milk is agitated, either by air or by more milk.
Unfortunately, situations where milk foaming occurs are
hard to avoid. First of all, almost all milking claws
are designed with an air bleed hole to allow the moving
of the milk towards the vacuum pumps by air pressure.
Under heavy milk flow conditions, this air bleed is
covered by fluid milk, thus creating foam as air enters
the hole and bubbles through the milk. The second prob-
lem occurs a little further down the line, whenever the
milk line enters a weigh jar or milk meter. If the con-
nection is made perpendicular to the side of the vessel,
the entering milk stream will impact either the vessel
walls or the milk collected inside. This turbulence
will create large amounts of foam that has a long per-
sistence time. The remedy for this is to introduce the
milk hose connection in such a manner that the connec-
tion is tangential to the vessel's inner diameter. If
this is done, the stream will adhere to the walls of the
vessel and.be conducted to the surface Of the milk below
with less foaming.

Accumulating the milk in a container having such
a tangential milk inlet reduces somewhat the problems of
two—phase flow and foaming. This means the milk can be
collected for measuring without causing any interference
to the flow Of air through the system, as the air can

pass out Of the vessel without bubbling through the milk.

12

The question that immediately arises, however, is how
much to accumulate? The accumulation could conceivably
range from a few ounces to a cow's entire production at
that milking. It was felt that a better milk-measuring
meter would result if the accumulation were kept as small
as feasible, reducing the bulkiness and related fragile-
ness that have been drawbacks of the weighjar. With this
in mind, it was decided to try to restrict the size of
the meter to about 2 liters of internal volume, with a
maximum milk accumulation of about 1/3 liter. This size
was arrived at in an arbitrary manner and was subject to
change.

Once the milk was accumulated, however, it was
necessary to get it out and to somehow measure the amount
of milk that left the vessel. In the early part of this
research, ideas were considered whereby some sort of
valve would be used to control the flow out of the vessel,
with the valve being controlled by conductive probes or a
float switch sensing the milk level. Due to problems
with valve sealing and cleaning, this approach was aban-
doned.

The next idea was to try to adapt some sort of
continuous flow device to measure the output from the
accumulation vessel. An Obvious candidate for this
device was a rotameter. This unit gives a continuous

visual output that varies with flowrate, and it was

13

felt that it would be possible to sense the position Of
the float in the rotameter by some electronic means Of
transduction.

One serious drawback to the rotameter existed,
however; that was the fact that a rotameter must be
oriented so that the flow being measured travels verti-
cally upward. NO way could be found to conveniently
direct the flow of milk in that direction and still
maintain compatibility with present milking systems.

It was felt that severe problems would be encountered
with milk flow reversal in the rotameter under certain
conditions. Since a rotameter cannot handle flow in a
reverse direction, errors of multiple-counting and high
total readings could result. There also would have been
problems with maintaining a single-phase flow up through
the rotameter during periods of fluctuating low output.
These problems made the rotameter unusable.

The idea, however, of some sort Of float which
would change position with increasing or decreasing flow
was felt to be worthy of further investigation. It was
hoped that the fewer moving parts Offered by such a
device would result in improved reliability and easier
cleaning. To achieve this goal would require that the
basic upward-flowing rotameter be transformed into a
device which operated in a downward—flowing mode.

Further examination of the rotameter revealed that

14

the constant factors on which the device could be cali-
brated were the weight of the indicator ball and the
specific gravity Of the fluid under measure. If an
empirical calibration was satisfactory for an applica-
tion, the weight of the indicator ball could be the
determining factor.

Therefore, to turn a rotameter-like device around
to operate in a downward-flowing mode, it would be neces-
sary to make some function other than the weight of the
indicator ball dependent on gravity. In other words, a
new constant had to be introduced into the rotameter. It
was felt that this would possibly be an application for
the electrodeless conductivity cell used by Gerrish
(1970). This cell was used to determine whether the
milk flow rate of a cow was over .05 pounds/minute or
not. It accomplished this by use of an orifice in the
milk line which would pass 0.5 pounds/minute under grav-
ity flow conditions and a bypass line around the orifice
which relieved any differential air pressure and any milk
in excess of 0.5 pounds per minute. The critical con-
cept gleaned from the electrodeless conductivity cell
was that a gravity-based, constant flow across an orifice
in a milking system could be achieved by use of a vacuum
bypass line around that orifice. Equally important was
the realization that the bypass in the milk-measuring

meter under consideration must pass only air and not

15

milk. Any milk which did not pass through the orifice
could not be measured and would therefore affect the
accuracy of the meter. These two concepts established
gravity-induced liquid flow as the standard phenomenon
around which the meter would operate. This phenomenon
would not predict the flow through the orifice, but
would ensure that vacuum fluctuations within the meter
or within the entire milking system would not disturb
the flow across the orifice.

Once the concepts were established concerning

the flow Of milk through the orifice, all that remained
was to measure that flow.) The main thrust of the
research at this point was to find a simple method of
producing an analog electrical signal that varied pro-
portionally with the flow rate through the orifice.
Many ideas were tried, some of which attempted to gener-
ate an electrical signal without an intermediate mechani-
cal step, and some of which used the motion of a mechani-
cal part of the meter to generate a signal when the milk
flow affected the mechanical part.

At some point during this process, the idea came
about to use a tapered float which rested in the orifice
under no-flow conditions, thereby closing off the orifice.
It was hoped that the float would rise out of the orifice
as sufficient milk accumulated around it to float it.

It was also hoped that as the flow rate increased, the

16

float would rise higher out of the orifice. If in fact
these things did prove true, it would seem possible to
measure the flow rate through the orifice by the vertical
displacement of the float.

Some preliminary test models were made, and a
configuration was found which worked satisfactorily.

The larger diameter of the tapered portion Of the float
rested in the orifice under no-flow conditions, and the
float would rise up with increasing flow, so that increas-
ingly smaller diameters Of the float were in the orifice.
This had the effect Of making the orifice larger. A
point would soon be reached where the flow coming into
the milk-measuring meter was equaled by the flow going
out of the orifice. The position Of the float would then
stabilize until the flow rate into the milk-measuring
meter changed, upsetting the equilibrium established by
the float and causing the float to move to a new posi-
tion of equilibrium.

These preliminary tests of the unit raised some
important questions. First Of all, did this unit have a
unique position of the float for every incremental change
in flow rate? If so, how small an increment of flow
could the milk-measuring meter resolve? Another import-
ant question was whether the displacement of the float
varied linearily with respect to the change in flow rate.

Put another way, were the flow rate and float displacement

17

related by a function such that a=bx, where a is the
change in flow rate and b is the change in displacement?
The preliminary answer to the first question was
arrived at qualitatively. Observation of the milk-
measuring meter in the laboratory under varying flow
rates indicated that a unique displacement of the float
did exist for any given flow rate as long as the flow
rate did not exceed the maximum capacity of the meter.
The final answer to this question would be obtained when
rigorous calibrations were undertaken. Calibration was
also the only method by which to answer the question of
whether the milk-measuring meter would act in a linear
fashion. An attempt was made to find a group of equa-
tions which would predict the performance of this valve,
but the attempt met with little success. Most of the
available work on flow prediction in control valves is
very theoretical and requires many assumptions. Even
the manufacturer's published predictions are sometimes
very inaccurate. Moore (1971), in speaking about this
deviation, states that "this deviation can be as much as
40%. Thus the precise mathematical formulation of the
optimum characteristics is not only prohibitive in terms
of complexity, but is also futile unless the actual char-
acteristic is known." With this advice in mind, it was

felt that the time necessary to select and evaluate a

18

predictive equation would be better spent by actually
calibrating the unit.

Before the milk-measuring meter could be built,
however, some method had to be found to transform the
mechanical motion of the float into an analog electronic
signal. The first criterion for this electronic inter-
face was sufficient resolution. Since it would even-
tually be desirable to digitize the output of the meter,
common sense decreed that the analog electronics should
be able to resolve a change in flow rate ten times
smaller than the smallest digitizing increment. The
second criterion, while not absolutely essential, was
certainly quite desirable: non-contact sensing. This '
would function without any electrical contact with the
milk being measured. Non-contact sensing would eliminate
problems with changing calibration due to a build-up of
contaminants and milk film on the conducting surfaces.

The criteria just summarized would seem to point
to some sort of electro-magnetic coupling. Several types
of circuits were tested, all of which employed a passive
element in the float which changed the inductance of a
stationary concentric coil as the float traveled up and
down within the coil. The circuit finally decided upon
was a Linear Variable Displacement Transformer (LVDT).

Herceg (1972) defines the LVDT:

19

Two identical secondaries, symmetrically spaced
from the primary, are connected externally in a
series-Opposing circuit. Motion of the non-
contacting magnetic core varies the mutual induc-
tance of each secondary to the primary, which
determines the voltage induced from the primary
to each secondary.
It was felt that this circuit would offer the best
results and would be the least troublesome to implement.
Also, as Herceg reports (1972), the LVDT has a linear
transfer function, meaning that the output of the Cir-
cuit is proportional to the movement Of the passive
element. Finally, the LVDT can be conditioned to pro-
duce as its output a varying DC voltage, highly suited
to digitization if desired.

Once all of the foregoing concepts were estab-
lished, a working model was constructed. Several iter-
ations incorporating minor changes were gone through
before a configuration was arrived at which was felt to
work reasonably well. The mechanical layout of this
meter, shown in Figure 1, consists of basically three
parts. These are the vacuum vessel, the receiver
compartment-float cup assembly, and the float itself.
The vacuum vessel is large enough to contain the
receiver compartment-float cup assembly and still allow
free air circulation around the assembly. It, like
almost all other parts in the prototype model, is made

of an acrylic plastic. Two hose connections are made

to the vacuum vessel, milk inlet and milk outlet.

20

 

 

 

 

 

 

 

 

 

LIP-J

 

 

 

 

 

 

 

 

 

n1 1: u
S n 0 HI...
N u m "rlrnullllHl’lhmm'l
m 1.. u n u. \UV
ch .. m "F... «iii».
EM. .5 "r----.il.....inl.__
Mm. :
DM E
On
C
MN 5
7
[I
m
Title}... .. m l
C»
T I
j m TI Wt. AI .UC m
1_N
I
u
_“
n .d:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L---
T.

i I

I

17....1

r

 

 

 

Figure l.--Overall Mechanical Layout.

 

.4 OUTLET

 

 

 

 

21

The receiver compartment-float cup assembly is
supported in the center of the vacuum vessel to insure
that the pressure differential across the receiver
compartment-float cup assembly, which contains the ori-
fice, is as close to zero as possible. This zero pres-
sure differential will give gravity flow across the
orifice, as was discussed earlier. The receiver compart-
ment uses the tangential inlet that has been shown neces-
sary to minimize the amount of milk foam created in the
vessel. The top part of the float rides in this compart—
ment. The float cup is attached to the bottom of the
receiver compartment and is the supporting structure for
the LVDT coils and the orifice plate. The orifice is
machined with a 90° contour and a diameter of 22 mm. The
LVDT coils are three coils of copper wire wound on the
outside of the float cup to facilitate electro—magnetic
coupling to the float. The coils are wound with AWG 36
copper wire and are 35 mm apart center-to-center. Each
of the coils are wound with 200 turns in a multi-layer
random pattern.

The float is constructed of machined acrylic
plastic. Detailed dimensions are given in Figure 2. An
early version attempted to use a float which had a total
diameter equal to the largest diameter of the tapered
section. This version had insufficient bouyancy to lift

itself properly, hence the development of the "mushroom"

22

‘TOF’VNEVV

 

 

 

 

 

 

 

 

 

 

.4— 610—»
n—lHJ—fi
I II, _
20.3 I
in}
78.7 I I
i 3“ FERRITE CORE
I I
was i I
i L..-J
ALL DIMENSIONS IN MM.

ALL HART?) ROUND.
CONCENTR IC.

SCALE 175 = 1

 

 

Figure 2.--Float Detail.

23

upper section. The tapered section of the float measures
51 mm by 24 mm, giving a taper of 14.5 degrees. The
float contains a slug of powdered iron which functions

to vary the inductive coupling between the three trans-
ducer coils. This slug is attached axially to a small
threaded rod, which extends through the top Of the float.
The rod permits vertical adjustment of the position Of
the slug in relationship to the float and the transducer
coils.

In action, the float and all of its supporting
structures work in a very straightforward manner. Milk
enters the receiver compartment through the tangential
inlet. It accumulates in the float cup and the receiver
compartment. As the level of milk rises, the float also
rises, retracting the tapered portion of the float and
opening the orifice. The level of milk will continue to
rise until the orifice is large enough to handle the
entire flow through the unit at that time. The vertical
displacement of the float therefore is proportional to
the flow rate of milk through the meter. This displace-
ment is sensed by the transducer coils and the attendant
electronic circuitry, which produces a DC output signal
proportional to the flow rate through the meter.

A block diagram of the electronic circuitry in
the milk-measuring meter is shown in Figure 3. The LVDT

circuit as used in this meter is an unusual adaptation

24

A

 

.mowcouuoon mo Emummfia xoonII.m ouawflm

 

 

 

mum—am

 

 

 

 

 

 

 

 

 

 

 

 

 

O. B 3.29 .>m

 

 

 

 

 

 

 

 

 

 

 

 

dwENDZOU

 

mwménp GA» > m

u_I0._.I>

 

 

 

 

 

mmow I:mN_NE.Z_

 

 

«an ><4amfin Na. \Ginpma muck/(.599. tut. ><Aama
wawmeZDOU 1! Nfl amt/Sou 1'. mfimthDOU :7. +9... mmPZDOU
mmkmz
30.5 '1
584mm _
D24 Run. > m.O I O
OnsrmNmuIn. .40rh<kHUXm_ .U< NIxmwN n
EUDQWZ/PE.

 

 

 

Q 00m 0. Zom._.><o

 

 

 

 

 

25

of a common concept. The LVDT is a widely used device

in industry, being an excellent means of measuring
straight-line mechanical displacements with a high degree
of accuracy. It is unusual, however, for an LVDT to have
both as large an inside diameter and as long a stroke as
this unit has. The expanded dimensions of the coils in
the meter were chosen to allow sufficient room around

the float for milk flow. A Daytronic 300D Measurement
Amplifier was used as both an excitation and readout

unit for the milk meter. It drives the center coil at
2.8 kHz and also functions as a phase-sensitive AC volt-
meter to tell the distance and direction the core is

from the null (center) position. This information is
available both from a front panel meter and as a 0 - ilO
VDC output signal, with 0.0 VDC corresponding to null
position. An important note, however, is that the milk
meter as constructed will use only one-half Of the avail-
able displacement range, specifically only one direction
from the null point. This will avoid any possible cross-
over distortion and slope change of the output curve due
to non-symmetry of the coils themselves.

Once the DC signal corresponding to flow rate is
obtained, the next step is to convert the signal into a
form that can be easily preserved. Since the goal of the
improved milk flow meter is to provide a readout of the

amount of milk produced, it was decided to integrate the

26

signal over time and convert it to digital form for
counting. TO accomplish this, an Intersil 8038 voltage-
to-frequency conversion IC is used. This circuit uses a
variable current source to linearily charge and discharge
a timing capacitor at a rate proportional to the flow
rate. A 8038 "squares up" the charge and discharge curve
of the capacitor with a comparator circuit, giving an
output which consists of a stream of 5V square waves
whose frequency is proportional to the flow rate. These
pulses, which are at digital logic levels, are then fed
to a series Of decade counters and light-emiting diode
(LED) displays which have a total capacity of 9999 counts.
This readout is not necessarily tied to any measurement
unit but can be calibrated for any reasonable unit by
manipulation of the voltage-to-frequency conversion pro-

CESS.

CALIBRATION AND TESTING

The task of calibrating the improved milk flow
meter was approached with two objectives in mind. The
first was to prove that the meter Operated in a linear
fashion, giving a constant change in output level for a
unit change in flow rate. The second objective was to
determine exactly what the calibration curve was and to
adjust the readout circuitry to reflect this curve.

Water was used in this calibration process
instead of milk. It had no noticeable effect on the
accuracy of the meter as opposed to the use of milk.

The specific gravity Of milk is 1.030-l.035 and the
viscosity of milk is 2.0 centipoise (Jenness, gt_al.,
1965). These figures closely approximate those of water,
eliminating the need for any artificial milk formula-
tions, etc., for early development work in this or simi-
lar environments. Water also has the advantage that it
does not spoil or otherwise make life difficult for those
using it.

The calibration of the meter itself and its
associated analog readout was accomplished by use of a
constant-head water-flow apparatus to smooth out fluc-

tuations in water line pressure, and a stainless steel
27

28

milk line valve (plug type) to control the flow into
the meter. This apparatus was capable of flow rates
in excess of 15 kilograms/minute, and as low as 5
grams/minute. The flow rate was adjusted until the
readout was at an 0.25 V interval over a 0.0-2.25 V
range. A stopwatch was used at each Of these settings
to measure the time required to fill a test container.
This container was weighed before testing to determine
the net mass of water it would hold. Dividing this
known mass by the filling time in seconds, and then
multiplying it by 60, gave the mass flow rate in
kilograms/minute. The data Obtained in this manner are
shown in Appendix B. Simple regression analysis was
applied to four replications of this test to produce
the calibration curve shown in Figure 4. The coeffi-
cient of correlation Of this curve gives an indication
of the good linearity of the analog portion of the
sensor under steady-state calibration conditions.

The calibration of the voltage-frequency con-
version section and the counting section was less diffi-
cult than the calibration of the analog section. The
flow rates developed during the analog calibration were
used to compute a corresponding frequency for each flow
rate. The voltage-frequency converter was fed voltages
at 0.25 V increments and adjusted for the corresponding

frequencies. The linearity of the voltage-frequency

29

maao>
m.~ m.H o.H m.o
a. lllllllll a. IIIIIIIII a. lllllllll ._. IIIIIIIII ._. IIIIIIIII
R
mzHD ous<auog<u u. .
mazHoa <a<a "A «
...
%
.....
a."
~aa.o u HoHuaaueeoo no .aauoo

\l
n“ .

.m>HDU coflumunHHnUII.v ousmfih

o'c---9-----‘o0--up----|-.--------.-opooo-uu+

O
O

o.e

. 25.: . OM

30

converter was accepted at the worst-case factory speci-
fications of 2.0%. The counter section was checked for
accuracy by feeding it a known number Of pulses and
verifying the count.

The first testing of the milk meter was done
in the laboratory. A total of 0.64 kg. of water was
poured through the meter at both slow (1.0 kg./min.)
and fast (2.5-3.0 kg./min.) rates. The data from this
test are shown in Appendix C.

The true test of the milk meter, however, came
in the milking parlor. The pulsating, two-phase flow
described earlier is very difficult to produce in the
laboratory, thus making parlor testing a necessity.
Parlor testing is usually not a desirable situation for
a prototype instrument, however, as the parlor is a
hostile physical and electronic environment. The large
amounts of water, the electrical noise from vacuum
pumps and other devices powered by electric motors,
and other related problems can induce errors in any
circuits that are not thoroughly shielded. Extensive
efforts were made to establish solid ground paths and
as much shielding as possible.

The parlor tests were conducted using a DHIA-
approved weigh jar as a control. The weight was read

from the jar in pounds and converted to kilograms to

31

allow comparison to the meter reading. These data can

be found in Appendix C.

ANALYSIS AND DISCUSSION OF RESULTS

The data collected from the laboratory testing
consisted Of ten repetitions from each of two tests.

Each of these repetitions was compared to the expected
vaer for the test and to the other nine replications
in the test. To perform this comparison, the sample
mean and the standard deviation of the sample for both
of the tests were computed.

The results of these computations showed that
the slow rate test had a mean of 0.58 kg. and a std.
deviation of 0.04 kg. The fast rate test showed a mean
of 0.66 kg. and a std. deviation of 0.03 kg. These
results, when compared to the expected value of 0.64 kg.,
show that under laboratory conditions the flow meter has
reasonable repeatability, but with some degree of non-
linearity. The data collected from the parlor testing,
however, were not as positive as that collected in the
laboratory. These data consisted Of 39 samples of paired
data, with each pair consisting Of a weigh jar reading
expressed in kilograms and a flow meter reading in kilo-
grams. Each of the flow meter readings was compared to

its related weigh jar reading, and then all of the pairs

32

33

of readings were related to each other. The statis-
tical measure chosen as most appropriate for this test
was the coefficient of correlation. It was chosen
because Of its ability to handle two random variables
at once, in this case the flow meter readings and the
weigh jar readings. The ideal results of this test
would have shown that these two random variables had a
position correlation coefficient of 1.00. If these
results were plotted, they would lie in a straight line
running at 45° to the axes. The results that were
obtained, however, are shown in Figure 5. Obviously,
they do not lie in a straight line. When the correla-
tion coefficient of thesedata points was computed,

it was found to be 0.128. It is very plain after exam-
ination of these results that the flow rate readings
have very little relationship to the actual weigh jar
readings.

A search for reasons which explained why the
flow meter does not work satisfactorily was only par-
tially successful. The fact that the flow meter per-
forms reasonably well in the laboratory under both
steady-state and intermittent flow conditions leads
one to consider problems found only in the milking
system as possible reasons for the malfunctioning of
the flow meter. One phenomena which was observed

during parlor testing was that foam sometimes collected

34

.mumo mCMUmoa Houoz 30HmII.m ousmflm

.Om .mmam:

o.oH o.~a o.w o.¢ o o
4 IIIIIIIII 4 IIIIIIIII a IIIIIIIII a IIIIIIIII +
o
o
o
p
o
o
o
p
. .
d o q
u
u
o
o o
. . 0m
0 a
o
o o
. .
. 4 o m
C O O D
o
o o
O C O O. D
o 00 b
O O. O. O
O O O O
C O O D
.. .. w o.NH
-
a
o
u
a
o
p
c
p
a.

.e<w onuz

35

under the "mushroom" top of the float. This lowered the

bouyancy Of the float and allowed it to "hang up" in

the float cup. The flow meter would then cease to record

flow until after the cow being milked was finished, when

the float cup would slowly drain out and free the float.
Other possible reasons for malfunction of the

flow meter could include vacuum fluctuations both in the

milking system and in the flow meter itself, air entrained

in the fluid milk, or impaired frequency response of the

float to milk flow transients.

CONCLUSIONS

1. The flow meter does not meet DHIA standards
for total milk weight. The correlation coefficient
between the actual milk weight and the weight recorded
by the flow meter does not fall within the range estab-
lished in the DHIA specifications.

2. The flow meter is compatible with the Opera-
tion of all currently-manufactured pipeline milking
systems. It does not introduce vacuum fluctuations or
disruptions into the milking system. Also, the flow
meter is designed to handle the pulsating flow caused
by lOOps and sags in the milk hose.

3. The flow meter is adaptable to clean-in-
place Operations. Some method will have to be found to
spray both the walls of the vacuum vessel and the inside
Of the receiver compartment, but this is within the
scepe of present day clean-in-place systems.

4. The flow meter is compatible with electronic
data recording systems. The data stored in the counter
units can be latched out to external circuits which will

interface it to the data recording circuits.

36

RECOMMENDATIONS FOR FUTURE RESEARCH

Work remains to be done on the configuration of
the receiver compartment-float cup assembly. It will
probably be necessary to increase the volume of the
receiver compartment by a factor of 3 or 4. It appears
to be important to allow sufficient room for the milk
stream to dissipate its kinetic energy after entering
the compartment.

Accuracy of the electronics could be improved by
the addition Of an 8-bit microprocessor and an 8-bit A-D
converter to replace the "charge-pump" integrated cir-
cuit with its capacitor leakage, and the discrete-chip
counter section. The addition of the microprocessor
would also greatly expand the potential usefulness of
this meter in the parlor for controlling a cow's detacher,
feed bowl cover, washer spray, recording her ID, etc.
This will truly make it part of a system, rather than

just an instrument.

37

APPENDICES

38

APPENDIX A

SCHEMATIC DIAGRAM OF ELECTRONICS

39

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.moacouuooam wo Emummfln oaumEosomII m ousmfim "MW 0mm Omm omm W...
Pwmmm i SOL. .5 g
a - _ 9m . a 9 n O u.
.mZOFmoEOmam 95%». m Iii. Zm
O. .59 < m
wt Q .220 0 >520 T A HUM—WW ~
, I m+
“I Imflfil IIHI m—l
uh _ 1o—
xmm u_ _ 9+
3. 96,— \+.
2 mu
3 T /
DOOm
_ Somu
Nmuxor fimuxoL

 

N.
mmow Z.
m

 

40

 

 

Pawn N
m

:
Omit Zm _
J>>xr
. o
a6?

N—
Y. 9.

r
E
N?

 

 

 

 

 

e—III'

 

———-l.n

 

 

 

 

 

 

 

 

 

 

 

 

APPENDIX B

CALIBRATION DATA

41

42

TABLE 1.--Ca1ibration Data.

 

 

Output Volts Seconds
Run # 1 2 3 4 5
0.25 72.0 72.0 84.5 73.5 75.5
0.50 39.0 41.0 42.0 40.0 40.5
0.75 27.0 29.0 29.0 27.5 28.2
1.00 20.0 21.0 21.5 20.5 20.8
1.25 17.0 17.5 17.5 17.0 17.3
1.50 14.5 15.0 15.0 14.0 14.6
1.75 13.0 13.0 13.0 13.0 13.0
2.00 11.5 11.5 11.5 11.5 11.5
2.25 10.0 10.0 10.5 10.0 10.1

 

Net calibration mass:

638 gm.

APPENDIX C

TESTING DATA

43

44

TABLE 2.--Laboratory Testing Data.

 

 

Slow Rate Fast Rate
Test No. (l kg./min.) (2.5-3.0 kg./minJ
l 0.63 0.70
2 0.64 0.70
3 0.56 0.65
4 0.58 0.68
5 0.59 0.65
6 0.59 0.61
7 0.59 0.69
8 0.59 0.63
9 0.53 0.68
10 0.54 0.64

 

Net mass:

0.64 kg.

45

TABLE 3.--Parlor Testing Data.

 

 

7.48

6.21

Weigh Jar, Kg. Flow Meter, Kg.
13.15 6.37
14.06 4.50

9.98 2.38
10.43 8.48
11.34 5.66
11.79 2.98
12.70 8.91
10.43 7.01
11.79 9.60
10.78 8.88

9.75 8.78

8.16 6.22
12.25 9.29
12.93 4.69

5.90 4.58

6.80 6.71

8.62 4.36
11.11 8.63

9.98 2.57
13.15 7.09
11.79 9.50
10.43 6.89
11.34 6.32
11.34 9.02

8.62 5.46

9.98 6.18

9.53 8.91
10.66 7.75
10.89 6.42

8.62 3.98

9.98 9.23
13.15 6.91
10.89 7.83
12.25 9.71
14.97 5.07
12.25 4.79
10.78 6.20
12.25 5.20

 

REFERENCES

46

REFERENCES

Bickert, W. G., J. B. Gerrish, and D. V. Armstrong, 1972.
Semi—automatic Milking in a Polygon Parlor: A
Simulation. Transactions of ASAE 15: 355-357,
360.

Gerrish, J. B., and W. G. Bickert, 1972. A Control Cir-
cuit and Transducer for an Automatic Milking
Machine Detacher. Transactions of ASAE 15: 189,
190, and 194.

Herceg, E. E., 1972. Handbook of Measurement and Con-
trol. Schaevitz Engineering, Pennsauken, N.J.
143 pp.

Irwin, James R., 1977. Scale, Weighing. The World Book
Encyclopedia 17: 142-143, Field Enterprises
Educational Corporation, Chicago, Ill.

Jenness, R., W. F. Shipe, and C. H. Whitnah, 1965.
Physical Properties of Milk. Chapter 8, pp. 339-
373, in B. H. Webb, ed., and A. H. Johnson, ed.,
Fundamentals of Dairy Chemistry. AVI Publishing
Company, Inc., Westport, Conn. 827 pp.

Moore, Ralph T., 1971. Flow Characteristics of Valves.
Chapter 5, pp. 119-139, in J. W. Hutchinson, ed.,
ISA Handbook of Control Valves. Instrument
Society of America, Pittsgurgh, PA. 310 pp.

 

Smith, J. W., and R. D. Plowman, 1969. Current Status
of Weighing, Sampling, and Testing Devices for
Use in the National Dairy Herd Improvement Pro-
gram. Journal of Dairy Science 52: 128-137.

47