Citation
The operational tuning of a spent grain dryer moisture control system

Material Information

Title:
The operational tuning of a spent grain dryer moisture control system
Creator:
Lucero, Wayne
Publication Date:
Language:
English
Physical Description:
viii, 64 leaves : illustrations ; 29 cm

Subjects

Subjects / Keywords:
Grain -- Drying ( lcsh )
Grain -- Drying ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaf 64).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Science, Electrical Engineering.
General Note:
Department of Electrical Engineering
Statement of Responsibility:
by Wayne Lucero.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
32480724 ( OCLC )
ocm32480724
Classification:
LD1190.E54 1994m .L83 ( lcc )

Full Text
THE OPERATIONAL TUNING OF A SPENT GRAIN DRYER MOISTURE
CONTROL SYSTEM
by
Wayne Lucero
B.S., University of Colorado at Denver
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Electrical Engineering
1994


This thesis for the Master of Science
degree by
Wayne Lucero
has been approved for the
Graduate School
by
Miloje S. Radenkovic
Hamid Z. Fardi


Lucero, Wayne (M.S., Electrical Engineering)
The Operational Tuning of a Spent Grain Dryer Moisture Control System
Thesis directed by Assistant Professor Miloje S. Radenkovic
ABSTRACT
This thesis details an investigation into the design, testing, and
implementation of a moisture control system for a rotary, steam tube spent
grain dryer at the Adolph Coors Company in Golden, Colorado. It includes
discussions on the entire spent grain drying process as well as the testing
and data analysis that accompany this type of R&D project. Various types of
instrumentation were used for this project and modifications to this
instrumentation were required throughout the process to obtain robust
performance. Documentation for these modifications is included in this
thesis along with discussions of numerous tests of these modifications.
Additionally, conclusions about the moisture control system and its direct
effect on areas such as safety and efficiency are presented in this thesis.
Further investigation is warranted into the potential usage of an automatic
moisture control system for drying spent grain.
This abstract accurately represents the content of the candidates thesis. I
recommend its publication.
Signed
Miloje S. Radenkovic
hi


ACKNOWLEDGMENTS
I would like to thank the Adolph Coors Company in Golden, Colorado,
for the opportunity to investigate their spent grain drying process and in
particular George Martin and Bryan Tway for their guidance and support
throughout this research.
IV


CONTENTS
Chapter
1. Introduction...........................................................1
1.1 Project Goals..........................!..............................1
1.2 Arrangement of the Thesis.............................................2
2. Literature Review.................................................... 4
3. Description of Drying Process..........................................9
4. Description of Existing System........................................12
5. Dryer Configuration................................................. 13
6. Equipment and Instrumentation....................................'....15
6.1 Mass Flow Meter.......................................................15
6.1.1 Description................................................. .....15
6.1.2 Calibration Procedure..............................................19
6.1.3 Mass Flow Meter Modifications......................................23
6.2 Moisture Sensor................................................... 24
6.2.1 Inlet Moisture Sensor..............................................25
6.2.2 Outlet Moisture Sensor........................................... 26
6.2.3 Calibration Procedure............................................ 27
6.2.4 Moisture Sensor Modifications......................................40
6.3 DCS (Distributed Control System)......................................41
7. Automatic Control System..............................................42
7.1 Description...........................................................42
7.1.1 Servo (Steam) Loop............................................... 42
v


CONTENTS
(cont.)
7.1.2 Regulator (Outlet Moisture) Loop............................43
7.1.3 Feedforward Loop............................................46
7.2 Tuning.......................................................46
7.3 Tests....................................................... 47
8. Economic Considerations...................................... 58
8.1 Proposed Control System #1:
0% Grain with Reduced Steam Usage.........................58
8.2 Proposed Control System #2:
8% Or Less Grain..........................................59
8.3 Proposed Control System #3:
6%-10% Grain..............................................60
9. Conclusions....................................................63
References.................................................... 64
VI


FIGURES
Figure
2.1. Step Response of a Feedback Only System...........................7
2.2. Step Response of a Feedback-plus-Feedforward System...............8
6.1. Mechanical Installation of Dryer #4 Mass Flow Meter...............16
6.2. Dryer #4 Mass Flow Meter Transmitter Panel........................18
6.3. Uncalibrated Instrument Response..................................31
6.4. Inlet Moisture Sensor Regression Plot of Instrument
vs. Laboratory Moisture Obtained on 2-11-94....................34
6.5. Inlet Moisture Sensor Regression Plot of Instrument
vs. Laboratory Moisture Obtained on 2-15-94....................35
6.6. Inlet Moisture Sensor Regression Plot of Instrument
vs. Laboratory Moisture Obtained on 2-22-94....................36
6.7. Outlet Moisture Sensor Regression Plot of Instrument
vs. Laboratory Moisture Obtained on 3-1-94.....................38
6.8. Outlet Moisture Sensor Regression Plot of Instrument
vs. Laboratory Moisture Obtained on 3-17-94....................39
7.1. System Block Diagram..............................................44
7.2 ISC (Inferential Smith Controller) Process Model Parameters.......45
VII


TABLES
Table
6.1. Dryer #4 Mass Flow Meter Calibration Data Table...................22
6.2. Inlet Moisture Calibration Data Obtained on 2-11-94.................34
6.3. Inlet Moisture Calibration Data Obtained on 2-15-94.................35
6.4. Inlet Moisture Calibration Data Obtained on 2-22-94.................36
6.5. Outlet Moisture Calibration Data Obtained on 3-1-94................38
6.6. Outlet Moisture Calibration Data Obtained on 3-17-94...............39
7.1. Spent Grain Dryer #4 Test Data Obtained on 3-1-94................51
7.2. Spent Grain Dryer #4 Test Data Obtained on 3-17-94............. 52
7.3. Spent Grain Dryer #4 Test Data Obtained on 3-22-94...............53
7.4. Spent Grain Dryer #4 Test Data Obtained on 3-23-94...............54
7.5. Spent Grain Dryer #4 Test Data Obtained on 5-20-94...............55
7.6. Spent Grain Dryer #4 Test Data Obtained on 5-25-94...............56
7.7. Spent Grain Dryer #4 Test Data Obtained on 6-1-94................57
viii


1. Introduction
1.1 Project Goals
The drying of cereal products has been a very challenging area in
process control for many years. However, with limited instrumentation
available for on-line measurements, little progress has been made in the
drying of cereal products. Another obstacle in successfully controlling outlet
moisture content is that the product transport lag or delay time of the system
is extremely large, rendering feedback techniques such as PID control
virtually useless. Today, with the availability of more sophisticated
controllers and on-line measurement devices, there is the potential for
significant improvements in how cereal products are dried and how much
control can be achieved on the outlet moisture content. At the Adolph Coors
Company in Golden, Colorado, a similar moisture control challenge exists in
the successful moisture control of spent grain.
The goal of this research project is to design, test, and implement an
automatic control system to regulate the outlet moisture content of spent
grain. It was determined by a group of Coors engineers, that grain with an
outlet moisture content of 8% produces higher quality grain for easier
pelletizing, increased safety of operation, and an overall more efficient
method of drying grain. Currently, there is no automatic control on the dryer
and the grain is over-dried to 0%, thus wasting valuable energy in the form
of steam, as well as evaporating the grains bound moisture that makes
1


pelletizing easier. Currently, the steam valve is open 100% at all times
regardless of the feed rate or moisture of the inlet grain. Theoretically, it
should take less steam to dry grain that contains less moisture than it does to
dry grain that contains a higher moisture.
1.2 Arrangement of the Thesis
Following is a brief outline of the thesis beginning with Chapter 2.
Chapter 2 contains a literature review of previously designed systems
and test results of similar high temperature grain dryers. A brief description
of each source cited as well as conclusions drawn from them are discussed
leading to the implementation and testing of the Spent Grain Dryer Moisture
Control System.
Chapter 3 contains an in depth description of the complete drying
process currently in place at the Adolph Coors Company. Each stage of the
drying process is discussed, explaining its purpose in the overall drying
process.
Chapter 4 describes the current manual control system being used to
dry spent grain and illustrates some of the disadvantages of such a system.
Chapter 5 provides an in depth look into the configuration of the dryer
itself, including a description of its internal structure and how it operates.
Chapter 6 gives an overview of the equipment used to implement the
Spent Grain Dryer Moisture Control System, including descriptions of the
DCS (Distributed Control System), moisture analyzers, and grain flow meter.
2


Discussions of the calibration procedures used for the instrumentation and
modifications made over the duration of the project are also presented in
Chapter 6.
Chapter 7 describes the entire Spent Grain Dryer Moisture Control
System and how the previously mentioned instrumentation works in the
system. Also, tuning procedures and test results are discussed in this
chapter.
Chapter 8 contains information on the economic considerations of the
system including benefits of an automatic control system.
Chapter 9 discusses conclusions drawn from the project, as well as a
list of recommendations for future consideration.
3


2. Literature Review
For many years, there had not been a great deal of consideration
given as to how product moisture control could be improved in the process
industry. But today it seems that any area that illustrates a potential of
cutting energy costs is being analyzed more closely (1). The area of product
moisture control is an area where significant energy, and revenue, are being
lost. For instance, in the case of spent grain, a significant amount of material
is being destroyed through over drying, and since the product is sold by
weight, this results in lost revenue. An example of the resulting lost revenue
follows:
Outlet Feed Rate = 4000 Ibs/hr
Actual Outlet Grain Moisture = 1%
Outlet Grain Moisture Specification = 5%
Moisture Leaving Dryer @ 5% = 200 Ibs/hr
Moisture Leaving Dryer @ 1% = 40 Ibs/hr
Difference (Solids Giveaway) 160 Ibs/hr
Operating 24 hours a day, 365 days a year, and selling for $90.00 per
ton translates into:
1RnM oRRdays 1ton $90-00 <,;Ro n7p nn/vr
160 hr 24 day 365 yr 2000lbs ton $63,072.00/yr
in lost revenue. Keep in mind that this is only the lost revenue as a result of
over-drying the grain, and does not include the substantial savings in energy
4


costs that would result from increased control. The control system discussed
in this reference is based upon the assumption that steam flow is directly
related to product moisture content. Therefore, if a control system could be
designed to control steam flow in the face of disturbances, then the
possibility of accurate product moisture control exists.
Isdale stresses the importance of some type of feedforward control
due to the large dead time of the system (2). Since he also was working with
limited on-line measurement devices, he also assumed that the steam flow
was directly related to the product moisture content. If the steam flow could
be controlled despite disturbances, then the product moisture content could
be controlled as well.
McFarlane emphasizes a feedback control system working together
with a feedforward control system (3). He explained that if there was some
way to measure the moisture content of the grain both entering and exiting
the dryer, then there is the potential of an improved automatic control
system. Figure 2.1 (page 7) illustrates the characteristic of a feedback only
system where Mj is the inlet moisture content, M0 is the outlet moisture
content, is the outlet target moisture content, and Tr is the residence time
of the grain in the dryer. Notice that the step response of the system is
delayed by one residence time of the dryer. In other words, with only the
outlet moisture content measurement available, the system does not react
fast enough to compensate for an increase in inlet moisture content until it is
5


too late, eventually resulting in wet grain exiting the dryer. On the other
hand, with an inlet moisture sensor and an outlet moisture sensor, a step
response could be obtained as indicated in Figure 2.2 (page 8). Note that
the effect of the feedforward term has placed the output m.c. (moisture
content) step symmetrically above and below the target. The control system
that I studied was very similar to Isdale's system, with the addition of inlet
feed rate as another variable used for feedforward control.
Finally, all three authors conclude that without accurate on-line
measurement devices, very little improvement on product moisture control is
possible. My objective is to support or disprove this theory based on my
research and implementation of such a system on a real world process.
6


M.
J.
o
Figure 2.1. Step Response of a Feedback Only System
Source: McFarlane, "Control of High-temperature Grain
Dryers", Measurement and Control, Vol.26, p. 208,
September 1993.
7


ta.c .
M.
l
0

T
r
------------------------------------------------------7
tune
Figure 2.2. Step Response of a Feedback-plus-Feedforward System
Source: McFarlane, N.J.B., "Control of High-temperature Grain
Dryers", Measurement and Control, Vol.26, p. 208,
September 1993.
8


3. Description of Drying Process
Spent grain is a by-product produced in the process of brewing beer.
Being the largest single site brewery in the world, the Adolph Coors
Company takes advantage of this large amount of by-product by drying it
and producing livestock pellets that are sold on site. But before spent grain
is transformed into pellets, it must proceed through many different stages of
drying.
Filter presses are one of the first devices used to separate the liquids
from the solids in the drying process. Filter presses are made up of several
cloth filters supported side by side. Liquid is forced through this series of
filters. Solids are captured in the filters and the liquid proceeds to the next
drying stage. After separating the filters from each other, wet grain drops
down into a large feed screw conveyor that moves the wet grain into a pump
that transfers the grain into holding tanks or a waiting truck.
These holding tanks (130 barrel capacity) are located on the 3rd floor
of K1 where there is a series of three holding tanks for each dryer.
Underneath each holding tank is a device called a Zenith Press that will be
discussed in detail in the next paragraph. Typically, two holding tanks and
two Zenith Presses are used during normal operation with the third tank and
press serving as a backup system in the event that problems would arise
with either of the other two systems. Additionally, when an immediate need
to remove wet grain from the brewery exists, the entire drying process is
9


bypassed and wet grain is pumped directly from the filter presses into a
waiting truck.
Zenith Presses are vertical devices that are gravity fed wet grain from
the previously mentioned holding tanks and are used to squeeze moisture
from the wet grain. They operate by moving wet grain through a vertically
oriented feed screw device that contains a series of circular channels that
decrease in width as the grain works its way to the bottom. The rotational
speed of the Zenith Presses is adjustable, depending on the brew schedule.
Centrifuges are another piece of equipment used in the drying
process. A centrifuge functions similarly to the spin cycle on a washing
machine, except that two tubs are used instead of one, one inside the other.
These tubs are horizontally oriented, approximately 8 long and 18" in
diameter, lie within each other, and are rotating at a very high speed in the
opposite direction. The centrifuges are supplied grain by the extracted liquid
from the Zenith Presses. Grain exits from one end of the centrifuge and is
sent into the dryer, while at the same time, liquids are exiting the opposite
end of the centrifuge and recycled to be mixed into another brew.
Also located on K1 2nd floor are additional pipes that supply both
reflux (dry spent grain) and wet spent hops. The first of the two most
common methods of supplying the dryer with wet grain are to have two
Zenith Presses producing wet grain and mixing this wet grain with reflux and
spent hops. This combination of solids is mixed together in a solids' mixer
10


that is located directly over the infeed chute to the dryer. Reflux is added to
the mixture because without it, the grain would be too wet to successfully dry
and would cause airveyor plugs on the outlet of the dryer. The other method
of supplying wet grain to the dryer is to have one Zenith Press running along
with one centrifuge. After moving through the dryer, the dry grain is
transported by an airveyor to storage silos. Dry grain is then moved back to
the pellet mills where it is turned into pellets and is either stored in silos or is
loaded into a waiting truck.
11


4. Description of Existing System
At the onset of this research, no automatic control system existed on
Spent Grain Dryer #4. The entire dryer system was solely under the manual
control of the operators and specialists. More specifically, the steam valve
was permanently opened 100% whenever the dryer was operating
regardless of inlet conditions. This meant that if there was a decrease in wet
grain being fed into the dryer, the valve remained at its maximum open
position, therefore wasting valuable steam. Additionally, if the dryer was to
run empty for any amount of time, the steam valve position was not
decreased, which also wasted a significant amount of steam. Apparently,
when the system was initially designed, it was not known exactly how much
steam was required to dry the grain to 0%. For example, it was unknown
whether the 100% valve position should correspond with 5000 Ibs/hr of
steam flow or if a fully opened steam valve should correspond with a steam
flow of 3000 Ibs/hr.
12


5. Dryer Configuration
Physically, the dryer is an 81 wide 120' long cylindrical drum with 132,
2" diameter steam tubes running the full length of the dryer. As the dryer
rotates at approximately 4-6 RPM, wet grain is fed into it at one end, and as a
result of the dryers slight inclination, rotation, and gravity, grain gradually
makes its way to the dryer discharge as dry product. As the dryer rotates, the
grain comes into contact with steam tubes that transfer heat into the grain.
Because of the dryers enormous size, the amount of heat that is generated
from the steam, and the proximity of three more identical dryers located
alongside Dryer #4, temperatures inside the room can approach 120
Fahrenheit. Unlike many rotary drum dryers, this dryer contains no strips or
paddles internally which assist in the mixing of the grain. Instead, it relies on
its axial transport tendencies to move grain through the dryer. Both the
rotational speed and angle of inclination are unadjustable constants in the
process.
Located at opposite ends of the dryer are different types of on-line
instrumentation that are essential to the success of the automatic control
system. At the inlet of the dryer is an impingement type solids' flow meter
that measures the velocity of the inlet grain flow as well as a dielectric
moisture sensor that measures the moisture content of the inlet grain. On
the outlet of the dryer is another moisture sensor that measures the moisture
content of the discharged grain. Descriptions of these instruments, along
13


with their calibration procedures, will be discussed in further detail in
Chapter 6.
14


6. Equipment and Instrumentation
6.1 Mass Flow Meter
6.1.1 Description
The sensor portion of the mass flow meter houses an LVDT (Linear
Variable Differential Transformer) and a support frame for the measuring
arm and sensing plate. The measuring arm connects the sensing plate to
the sensor and mechanically transmits the deflection of the sensing plate to
the LVDT. The amount of sensing plate deflection, which is proportional to
the mass flow, is converted to a voltage by the LVDT. This voltage is
changed to a pulsed frequency modulation signal (PFM) by a voltage to
frequency converter within the preamplifier. From the preamplifier, the
signal is transmitted to the flow transmitter through a two conductor cable.
The transmitter computes, displays and outputs the mass flow signal
generated by the sensor. This transmitter is used to access the programs for
specific input of process parameters used in calculating the mass flow. The
transmitter also provides the power supply for the preamplifier and LVDT
through the same cable the signals are transmitted to the transmitter.
Additionally, a 4..20mA signal representing the mass flow is transmitted to
the DCS (Distributed Control System) for use in the control algorithm.
Figure 6.1 (page16) illustrates the mechanical installation of the
sensing plate, similar to the one implemented on Dryer #4.
15


Figure 6.1. Mechanical Installation of Dryer #4 Mass Flow Meter
From: Endress and Hauser, Granumet DE10 and DME270 Impact Weigher
Installation and Operations Manual, No. IM 003G/01/ae/04,93.
Reprinted with permission.
16


The impact force for the flow at non-reference conditions can be calculated
from the parameters shown in Figure 6.1 (page 16) so that the proper
measuring range can be selected. The following equation applies for
determining the measuring force range:
FM = qnWh(0.025) sin a sin y (6.1)
Where:
Fm = Horizontal force on the sensing plate measured in Newtons
qm = h
h = Height of fall on inches (initial velocity should be 0 if possible)
a = Angle of sensing plate to the horizontal plane (sensing plate
angle)
y = Angle between impact angle and falling material (sensing angle)
The operational parameters and display quantities for the transmitter
are arranged in the form of a 10 by 10 matrix. The individual display or input
positions can be selected through a code and can be read on the 4-line LCD
display of the transmitter panel shown in Figure 6.2 (page 18). The code
consists of a row number (H, horizontal) and a column number (V, vertical).
V and H coordinates are expressed in numerical values from 0 to 9. VO, HO
is considered Home Position and is where the normal operation information
is displayed.
17


LEO, Counter Rslay 1
LEO, Alarm flalayo
Inquiry LEO
Rackbuo intarfaca
LEO, Syatam Error
Taat Points 0/4-20 mA
Analog Output 1
Taat Points 0/4-20 mA
Analog Output 2
Homo Diagnostics Lsft Down Up night Entar
Figure 6.2. Dryer #4 Mass Flow Meter Transmitter Panel
From: Endress and Hauser, Granumet DE10 and DME270 Impact Weigher
Installation and Operations Manual, No. IM 003G/01/ae/04,93.
Reprinted with permission.
18


6.1.2 Calibration Procedure (4)
After entering the installation parameters, the transmitter contained all
the information necessary to interpret the force transmitted to the impact
plate by the material flow as a theoretical flow rate. However, because of the
unpredictable nonlinearities inherent to the measuring principle of impact
weighing, it was necessary to perform a calibration over the entire
measuring range. To do this, a series of comparative test weighings was
performed where a known quantity of grain was conveyed through the
sensor. The value indicated by the sensor was then compared to the actual
weight (known as a reference weight) of the conveyed grain to calculate a
calibration curve for that particular flow rate. Since multiple flow rates are
common at the inlet of the dryer, it was recommended by the manufacturer
that one calibration point be set at three separate flow rates of 30%, 50%,
and 80% of the maximum flow rate of the application. Table 6.1 (page 22) is
the table used to record calibration values during the calibration procedure.
Following is a detailed calibration procedure followed to calibrate the
Mass Flow Meter:
Test Weighing at 50% Flow Rate
1. Go to V5 H1, Calibration Point. Enter 0. This erases any
information that may have been previously input and resets the
point to number 1.
2. Go to V5 H2, Calibration Start/Stop. Press the E key to start the
flow at 50%. A flashing warning LED indicates that the high
resolution calibration counter is activated.
19


3. After the complete test sample has passed the sensor, press the
"E key again to stop the counter. Record the value displayed in
line 3 of the display in the appropriate column of Table 6.1 (page
22). Note that any value greater than 999 lbs is displayed in tons.
4. Go to V5 H3, Reference Weight. Press the E key, then enter the
known weight of the test sample. Any value greater than 999.9 lbs
must be entered in tons and any value less than 0.5 tons must be
entered in lbs. After the correct value is entered, press the E key
to accept the value. Record this value in the correct column of
Table 6.1 (page 22).
5. Go to V5 H4, Correction Factor/Calibration Factor. The value
initially displayed here is the correction factor. This value is
automatically calculated by the flow transmitter by dividing the
reference weight (V5 H3) by the calibration counter value (V5 H2).
Record this value in the appropriate column in Table 6.1 (page
22). Note that if at any time during the calibration procedure the
correction factor is 0.99 to 1.01, the measurement error is less than
1% and the correction factor should not be accepted. To accept
this value as the correction factor, press the E key twice. When
the correction factor is accepted the text will change from
Correction Factor to Calibration Factor and the value displayed
is the new calibration factor. Record this value in the appropriate
column of Table 6.1 (page 22). The calibration factor is
automatically calculated by the flow transmitter as follows:
New Cal. Factor = Old Cal. Factor Correction Factor (6.2)
6. Record the values in positions V5 H5 thru V5 H7 in the appropriate
columns in Table 6.1 (page 22).
7. Perform steps 2 thru 5 at least three times total for each flow rate to
achieve higher accuracy and to ensure that the measurement is
repeatable.
Test Weighing at 80% Flow Rate
1. Go to V5 H1, Calibration Point Number. Enter 2.
2. Perform calibration procedures as described in steps 2 thru 6
20


above except set the flow rate at 80% of the maximum.
Test Weighing at 30% Flow Rate
1. Go to V5 H1, Calibration Point Number. Enter 2.
2. Perform calibration procedures as described in steps 2 thru 6
above except set the flow rate at 30% of the maximum.
After carrying out this procedure, mechanical installation flaws were
discovered with the sensor that caused the initial calibration figures to be
inaccurate and unusable. As a result of these flaws, modifications to the
mechanical installation were required and they are described in the
following section.
21



Calibration Point # (V5H1) Calibration Counter (V5H2) Reference Value (V5 H3) Correction Factor (V5 H4) Calibration Factor (V5H5) Average Flow Rate (V5 H5) Average PFM (V5 H6) Peak PFM (V5H7)


















Table 6.1. Dryer #4 Mass Flow Meter Calibration Data Table
22


6.1.3 Mass Flow Meter Modifications
On April 1, 1994, the following modifications were done on the infeed
conveyor, infeed chute, and impact plate:
Flighting of the infeed conveyor was decreased (cut)
The inlet chute above the impact plate was decreased (cut)
A liner was fabricated and fastened to the impact plate
Along with these modifications, a limit switch activated alarm was installed
above the impact plate housing to alert operators and specialists of a grain
plug at the flow meter. This included installing a limit switch, an alarm horn,
relays, and running the necessary wires.
After these modifications, an attempt to calibrate the grain flow meter
was again attempted on July 5, 1994, with little success. This time, based on
comments from the manufacturer's representative, the problem pointed to a
faulty transmitter that was shipped to the manufacturer for repair and
returned to Coors on July 14, 1994, after the manufacturer's engineers and
technicians could find nothing wrong with the instrument. Another
calibration attempt was made on July 19, 1994, after it was discovered that
the transport arrest bracket on the vertical damper had not been moved to
the free position which caused the shaft and measuring plate to always be
stationary (0 deflection). After carrying out the entire start up process again
including leveling the sensor, adjusting the LVDT, and filling the horizontal
damper with silicon oil, I was able to successfully calibrate the instrument
23


and it seems to be operating adequately.
6.2 Moisture Sensor
Located on the outlet of the dryer is a dielectric moisture sensor that
measures moisture content of the exiting grain and displays this value on a
digital display as well as transmits this value to the DCS by way of a 4..20mA
signal. Basically, the sensor operates as follows:
1. A sample of grain is collected on the sensors collection tray
(collection time)
2. A measurement is performed to determine the grains moisture
content (measurement time)
3. The moisture content value is displayed on the processor and sent
to the DCS in the form of percentage by weight
4. The measured sample is blown off of the tray (purge time) and the
measurement sequence is repeated
When this research initially began, both the inlet moisture sensor and the
outlet moisture sensor were identical models that required manual relay time
settings for the collection time, measurement time, and purge time. Since
then they have both been replaced with upgraded units that have an
alphanumeric keypad and digital display that are used to set only the
collection time and purge time. There is no longer a need to set the
measurement time on the newer models because the processor
automatically performs the measurement at the appropriate time. Additional
modifications of each unit will be discussed further in Section 6.2.4.
24


6.2.1 Inlet Moisture Sensor
As previously stated, originally, the type of sensor on the inlet of the
dryer was identical to the sensor on the outlet of the dryer. After installing
the original inlet moisture sensor, the following significant observation was
made:
The air purge system was not clearing the sample tray of the wet
grain, thus causing the sensor to repeatedly measure the same
sample.
In an attempt to remedy this problem, an external air tank was hooked up to
the purging system to provide a larger air blast to clear the tray of wet
material, but this did not solve the problem. As a result of the air purging
problems, it was unable to determine if the sensor was making accurate
moisture measurements on the 40-70% grain. On April 15,1994, the sensor
was removed from the inlet chute and returned to the manufacturer for their
evaluation of the instrument that could be contributing to the erratic moisture
readings. They discovered some physical damage to the sample tray and
requested that a large sample of wet grain be sent to them so that they could
conduct tests to determine the feasibility of wet grain measurement before
developing a better sampling mechanism. The manufacturer determined
that the instrument was capable of measuring wet grain accurately so they
proceeded to design a sensor that collected grain similar to the earlier inlet
moisture sensor, but instead of using an air purge system to clear the tray of
25


wet grain, they would design a pneumatic cylinder to drive a wiper plate
across the sensor to push product off. On August 10, 1994, Coors received
the modified inlet sensor but one of the required specifications was not met.
Because of obstructions near the inlet chute area, longer standoffs were
required to mount the moisture sensor. Longer standoffs were fabricated but
before the moisture sensor could be installed, new cables had to be sent
from the manufacturer because the longer standoffs caused the two existing
white coaxial cables running from the electronics enclosure to the dielectric
plate to be too short. After receiving and connecting the new cables, the
moisture sensor was installed on August 19, 1994, and is performing
satisfactorily.
6.2.2 Outlet Moisture Sensor
Before the beginning of this research, the outlet moisture sensor had
been installed and somewhat calibrated. The purpose of the outlet moisture
sensor is to provide information about the outlet moisture content of the
exiting grain. It works identically to that of the initially installed inlet moisture
sensor, in that it collects a sample, measures the sample, then blows the
measured sample off the tray before repeating the process. On
approximately June 10, 1994, the outlet moisture sensor was returned to the
manufacturer and replaced with an upgraded model. The new model
sensor was installed and has worked accurately on all tests performed as of
July 5, 1994.
26


6.2.3 Calibration Procedure (5)
Following is the procedure used to calibrate both the inlet and outlet
moisture sensors as well as results of calibrations performed on both
moisture sensors.
Pre-zero
The purpose of pre-zero is to remove the influence of the surrounding
environment upon the sensor, e.g., a metal beam located three inches from
the face of the sensor, a plastic window between the sensor and the grain or
in this particular application, the large amount of metal around the sensor
tray. If this residual dielectric were not negated, it would cause problems
when calibrating. When the instrument is pre-zeroed, a single dielectric
measurement is made and stored in memory and this pre-zero value is
subsequently subtracted from all future readings. The basic moisture
algorithm used in the sensor is shown below:
Moisture = Dielectric Span [Dielectric Pre-zero]
+ Dielectric Zero (6.3)
Pre-zero was performed on the instrument as follows:
1. All grain was removed form the sample tray
2. I pressed 6, then F, then ENT
After completing the pre-zero procedure on the sensor, l then proceeded to
standardize the instrument.
27


Standardization
The purpose of standardization was to provide uniformity of
calibration between instruments, and to provide a repeatable reference. The
standardization value is used as a secondary span coefficient, and may be
considered as a scaling factor (S.F.) in the following equation:
Moisture = S.F. Dielectric Span [Dielectric Pre-zero]
+ Dielectric Zero (6.4)
Due to manufacturing and component tolerances, no two sensors will be
identical, thus the scaling factor or standardization value will compensate for
these differences. During standardization, Dielectric Span is forced to one
(1), and Dielectric Zero is forced to zero (0) thus, Eq. (6.4) becomes:
Moisture = S.F. (Dielectric Pre-zero) (6.5)
Transposing Eq. (6.5) gives:
Moisture
S.F. =
(6.6)
(Dielectric Pre-zero)
If a known reference sample was available, Standardization could have
been done by entering this value and the processor would have calculated
S.F. from Eq. (6.6). However, since a phenolic plate was provided by the
manufacturer, I simply placed this plate in the tray and then entered its target
value of 25 into the processor allowing the instrument to calculate an S.F.
from this information. The following steps illustrate how I carried out the
procedure:
28


1. I placed the black phenolic plate in the sample tray
2. I entered 7F\ then ENT
3. I entered 25, then ENT
After carrying out the Pre-zero and Standardization procedures, I then
proceeded to Basic Calibration.
Basic Calibration
The moisture sensors used in this research measure the dielectric
constant of the grain using a patented Resonant Frequency Technique
where dielectric measurements are performed at 10 ms. intervals. Two
reference capacitors are switched in place of the sensor once per second for
10ms. each. These are termed the high and low references that produce
high and low reference frequencies. The antenna frequency and reference
frequencies are processed to produce a very stable calibration signal
referred to as the Raw Dielectric. Since the Raw Dielectric is a function of
sensor capacitance combined with product capacitance, the first step in
basic calibration is to isolate the product capacitance. This is achieved with
the Pre-zero function performed earlier. Recall that when the instrument was
Pre-zeroed, the dielectric of the empty sensor was measured and stored in
non-volatile memory. This value is subsequently subtracted from all future
measurements and is reflected in the following equation:
Product Dielectric = Raw Dielectric Pre-zero (6.7)
The product dielectric is proportional to moisture content and can be scaled
29


to give exact moisture by applying a multiplier (Span) and a bias (Zero)
reflected below in Eq. (6.8).
Moisture Content = Span Product Dielectric + Zero (6.8)
Figure 6.3 (page 31) shows a typical instrument response to moisture before
calibration. A linear function is obtained of the form y = (mx + b) or in this
case:
Instrument = (Slope Moisture + Intercept)
(6.9)
where:
Y Y
Slope =x22_ X~1 and,
Intercept = Y0
The positive intercept is a result of the dry product having some residual
dielectric. To obtain a calibrated response, it was necessary to normalize
Eq. (6.9) by dividing both sides by the Slope, then subtracting the Intercept
resulting in the following equation:
Instrument Intercept
Moisture =- slope
Comparing Eq. (6.10) to Eq. (6.8) shows that:
(6.10)
Span =
_J___
Slope
and
Zero
-Intercept
Slope
30


Instrument Moisture (%)
Laboratory Moisture (%)
Figure 6.3. Uncalibrated Instrument Response
31


Note that instrument Span adjusts the slope of the graph and
instrument Zero applies a constant offset to negate Intercept. If the graph of
Figure 6.3 (page 31) were obtained with a known Span and Zero, the
desired Span and Zero could be calculated from:
Required Span = n slope ^ (6.11)
Required Zero = ^M!M|^tercept (6.12)
In calibrating both the inlet moisture sensor and the outlet moisture
sensor, I was required to obtain a graph similar to that shown in Figure 6.3
(page 31). For the Span and Zero calculations, I used the bench moisture
meter located in the dryer room to determine actual moisture content since
bench moisture measurements were very close to laboratory moisture
measurements. Table 6.2 (page 34) shows the measured data taken on
February 11, 1994, on the inlet moisture sensor. The table shows the
sample number, the time of each sample collected, the Span and Zero
settings at the time of the readings, the reading on the moisture sensor, and
the results from the laboratory measurements. Figure 6.4 (page 34) shows a
graphical representation of Table 6.2 (page 34) values with the relationship
between the Laboratory Moisture and the Instrument Moisture. After using
Eq. (6.11) and Eq. (6.12) to calculate new values for Span and Zero based
on the Slope and Intercept of Figure 6.4 (page 34), another set of data was
32


obtained on February 15, 1994, shown in Table 6.3 (page 35) and shown
graphically in Figure 6.5 (page 35). Again, after calculating the new values
for Span and Zero, I obtained the data on February 22, 1994, shown in
Table 6.4 (page 36) and shown graphically in Figure 6.6 (page 36). Shortly
after this calibration attempt, the inlet moisture sensor was returned to the
manufacturer for modifications.
33


Sample # Time Instrument Moisture (%) Laboratory Moisture (%> Span/Zero
1 9:00 AM 46 43.67 0.06/0.00

2 9:05 AM 45.6 43.84 0.06/0.00

3 9:10 AM 45.3 41 0.06/0.00

4 9:15 AM 45.6 42.6 0.06/0.00

5 9:20 AM 44.8 44.05 0.06/0.00
Table 6.2. Inlet Moisture Calibration Data Obtained on 2-11-94
Laboratory Moisture (%)
Figure 6.4. Inlet Moisture Sensor Regression Plot of Instrument vs.
Laboratory Moisture Obtained on 2-11-94
34


Sample # Time Instrument Moisture (%) Laboratory Moisture (%) Span/Zero
1 8:55 AM 53.2 44.26 0.01/-60
2 9:05 AM 53.8 41.5 0.01/-60
3 9:15 AM 70.8 41.91 0.01/-60
4 9:40 AM 56 42.93 0.01/-60
5 9:45 AM 54.1 42.91 0.01/-60
6 9:50 AM 54.5 41.63 0.01/-60
7 10:05 AM 52.5 24.57 0.01/-60
8 10:10 AM 52.5 24.93 0.01/-60
9 10:15 AM 52.4 24.79 0.01/-60
Table 6.3. Inlet Moisture Calibration Data Obtained on 2-15-94
CT'
d>
<0
O
Laboratory Moisture (%)
Figure 6.5. Inlet Moisture Sensor Regression Plot of Instrument vs.
Laboratory Moisture Obtained on 2-15-94
35


Sample # Time instrument Moisture (%) Laboratory Moisture (%> Span/Zero
1 8:30 AM 34.4 50.1 0.01 /-70
2 8:35 AM 34.8 49.8 0.01/-70
3 8:40 AM 34.7 49.67 0.01/-70
4 8:45 AM 34.4 49.6 0.01/-70
5 10:15 AM 34.8 47.57 0.01/-70
6 10:20 AM 34.8 48.69 0.01/-70
7 10:25 AM 35.3 51.45 0.01/-70
8 10:30 AM 35 52.11 0.01/-70
9 11:30 AM 32.8 42.46 0.01/-70
10 11:35 AM 34.1 48.55 0.01/-70
11 11:40 AM 33.9 48.11 0.01/-70
12 11:45 AM 35.6 51.45 0.01/-70
Table 6.4. Inlet Moisture Calibration Data Obtained on 2-22-94
Laboratory Moisture (%)
Figure 6.6. Inlet Moisture Sensor Regression Plot of Instrument vs.
Laboratory Moisture Obtained on 2-22-94
36


The first data obtained on the outlet moisture sensor was obtained on
March 1, 1994, and is shown in Table 6.5 (page 38) and graphically in
Figure 6.7 (page 38). From this information, new values of Span and Zero
were calculated and entered into the processor and on March 17, 1994, data
was compiled shown in Table 6.6 (page 39) and shown graphically in Figure
6.8 (page 39). At the conclusion of this calibration attempt, the outlet
moisture sensor was returned to the manufacturer in exchange for a newer
model with upgraded features and improved performance.
37


Sample # Time Instrument Moisture (%) Laboratory Moisture (%) Span/Zero
1 8:15 AM 1.2 1.2 1.03/-16.24
2 9:00 AM 0.9 3.6 1.03/-16.24
3 9:30 AM 3.4 7.2 1.03/-16.24
4 10:15 AM 5.5 10 1.03/-16.24
5 10:45 AM 7.3 12.6 1.03/-16.24
6 11:20 AM 8.9 12.4 1.03/-16.24
Table 6.5. Outlet Moisture Calibration Data Obtained on 3-1-94
a>
k.
3
(0
o
2
c

E
3
(0
c
Laboratory Moisture (%)
Figure 6.7. Outlet Moisture Sensor Regression Plot of Instrument vs.
Laboratory Moisture Obtained on 3-1-94
38


Sample # Time Instrument Moisture (%) Laboratory Moisture (%> Span/Zero
1 8:00 AM 0.2 0.73 0.67/-11.20
2 9:45 AM 0.4 2.21 0.67/-11.20
3 10:00 AM 0.7 2.24 0.67/-11.20
4 10:30 AM 1 2.93 0.67/-11.20
5 11:00 AM 1.3 3.95 0.67/-11.20
6 11:45 AM 2.8 5.38 0.67/-11.20
Table 6.6. Outlet Moisture Calibration Data Obtained on 3-17-94
a)
3
**
(0
o
c
a>
E
3
(0
c
Laboratory Moisture (%)
Figure 6.8. Outlet Moisture Sensor Regression Plot of Instrument vs.
Laboratory Moisture Obtained on 3-17-94
39


Entering Span and Zero
To change Span and Zero settings in the sensor processor, the
following steps were taken (from measurement mode):
1. I pressed FFP, the ENT
2. I pressed F, then entered new Span, then ENT
3. I pressed F( then entered new Zero, then ENT
6.2.4 Moisture Sensor Modifications
As mentioned earlier, both the inlet moisture sensor and the outlet
moisture sensor have undergone significant modifications since the
beginning of this research.
Inlet Moisture Sensor
On August 19, 1994, the modified inlet moisture sensor was installed
and seemed to be working satisfactorily even though there was insufficient
time for me to carry out any calibration procedures.
Outlet Moisture Sensor
The newly designed outlet moisture sensor has undergone significant
changes since the inception of this research. Following is an up-to-date list
of the modifications done to the outlet moisture sensor:
1. The entire sensor and processor have been replaced with a newer
model.
2. A baffle has been installed to the end of the sample tray to insure a
full tray of material during the measurement cycle, as well as to
40


prevent grain from being blown out of the dryer onto the floor
during the purge time of the cycle.
6.3 DCS (Distributed Control System)
At the controls of this system is a function.block based programmable
DCS. Residing in this controller is a Smart Module, Analog Inputs, Analog
Outputs, Digital Inputs, and Digital Outputs. It operates by executing function
block software which can be modified on a PC and downloaded to the
module or can be directly modified and monitored on-line using a hand-held
module. The controller has a digital display face plate that, depending on
which loop is active, displays: steam flow, valve position, outlet moisture
and set point. The controller can also operate in three different modes:
manual, auto, and cascade. Lastly, this DCS is a stand-alone independent
controller with no requirements for external I/O modules.
41


7. Automatic Control System
The primary goal of this research was to design, test, and implement
an automatic moisture control system to regulate outlet moisture content of
spent grain. At the beginning of this research, the initial design of the system
was in place but had not been tested or tuned and I've included a
description of this entire control system design.
7.1 Description
Figure 7.1 (page 44) shows a block diagram of the control system
design. Essentially, there are three separate loops working together shown
in this diagram, namely: The Servo (Steam) Loop, The Regulator (Outlet
Moisture) Loop, and The Feedforward Loop. The control philosophy of each
of these loops will now be discussed in detail.
7.1.1 Servo (Steam) Loop
The functions of the Servo (Steam) Loop are as follows:
Receives a set point generated by the feedforward and feedback
loops and its control output sets the.steam valve position.
Reacts immediately to disturbances in the steam and/or
condensate system, before disturbances affect outlet moisture
content.
42


7.1.2 Regulator (Outlet Moisture) Loop
The functions of the Regulator (Outlet Moisture) Loop are as follows:
Generates the set point for the Servo Loop.
Utilizes an ISC (Inferential Smith Controller) instead of a PID
Controller because of the large process lag or dead time. The
Smith Controller models the system with three parameters:
Gain (K) =
A Outlet Moisture
A Steam Flow Rate
, 4 2 A Outlet Moisture
Lag Time (L) = o r----------------
3 A Steam Flow Rate
Dead Time (D.) = the time for Outlet Moisture to
respond to a change in Steam Flow.
These general parameters are illustrated in Figure 7.2 (page 45). The Gain,
Lag Time, and Dead Time are all time varying functions of the grain flow
rate. Another term for this type of control is Self-Tuning Control.
43


Inlet


K =
aco C2 q
D = t1 tfl
L = t2 t,
Figure 7.2. ISC Process Model Parameters
From: Bailey DCS Function Code Manual, September 15, 1992
45


7.1.3 Feedforward Loop
The functions of the Feedforward Loop are as follows:
Inlet grain flow rate and inlet moisture content are multiplied
together to generate the moisture load to the dryer.
The moisture load is added to the set point generated feedback
signal from the Servo Loop.
A lag model is used with the feedforward signal and this lag
model utilizes the following 3 tuning parameters:
A Outlet Moisture
GainFeedforward (KFeedforward) =A Moisture Load
... 2 A Outlet Moisture
Lsg TimeFeedforward (L) = w ~ ; 7 \ ~r
6 A Moisture Load
Dead TimeFeedforward (D) = the time for Outlet Moisture to
respond to a change in Moisture
Load.
7.2 Tuning
As stated earlier, the control system design had already been
completed when I started this research, but no tests had been conducted to
determine the proper tuning parameters for neither the Regulator Loop nor
the Feedforward Loop. Therefore, it was decided that to determine the
tuning parameters for the Regulator Loop, steam flow would be adjusted to
reach a steady state condition of a constant outlet moisture content and then
a step change would be initiated by changing the steam flow. Unchanging
inlet conditions of both inlet moisture content and grain flow rate would be
46


required to successfully determine tuning parameters. The objective was to
obtain an ISC process response similar to that of Figure 7.2 (page 45) where
the Controller Output (CO) is steam valve position and the Process Variable
(PV) is outlet moisture content. After completing the test to determine the
tuning parameters for the ISC, this test was to be repeated to determine if the
tuning parameters were different for different inlet conditions. Therefore, one
can see the importance of properly calibrated instrumentation to obtain a
robust moisture control system.
7.3 Tests
Numerous tests were conducted throughout this research to
determine the effects of decreased steam flow. Table 7.1 (page 51) contains
data taken on March 1, 1994, where steam flow was decreased from
4400lbs/hr to 1800lbs/hr and measurements were taken every 15 minutes. It
includes date or time, inlet moisture percentage, outlet moisture percentage,
steam flow rate, steam pressure, condensate pressure, grain temperature,
stack temperature, and Zenith Press speeds. Notice that the outlet moisture
content began to significantly change after about an hour and a half after
steam flow was decreased. This test provided information on how long it
took for the outlet moisture content to start changing as a result of decreased
steam flow, as well as illustrate the effects on grain temperature and other
measurable parameters.
Tables 7.2-7.6 (pages 52-56) show the results conducted on March
47


17, 22, and 23, 1994 as well as tests conducted on May 20, and 25, 1994.
Each of these tests was initiated differently to test different characteristics of
the system.
On June 1, 1994, a test was conducted to determine how well the
feedback portion of the system would perform in the face of changing inlet
conditions (both feed rate and moisture content). Software on the DCS was
set up to perform in the following manner:
1. Set point for steam flow was set at 3000lbs/hr.
2. Tracking value for valve position was set at 100%.
The objective was for the controller to monitor the steam flow and to regulate
the valve position to keep 3000lbs/hr of steam flow. At the same time, the
controller was to monitor the outlet moisture signal from the outlet moisture
sensor and transfer into tracking mode if the outlet moisture reached 2% or
greater. While in the tracking mode, the valve was to open up 100% until the
2% criterion was met. After which point, the controller would start regulating
valve position based on its set point. Prior to the start of the test, at 8:00AM
on June 1, 1994, the inlet conditions were recorded as follows:
1. Two Zenith Presses operating at 10 RPM each combined with
reflux and spent hops were feeding the dryer.
2. Inlet grain moisture (measured on the bench) was measured to be
between 69% and 71%.
48


The data obtained during this test is shown in Table 7.7 (page 57). With the
set point at 3000lbs/hr from 8:00AM to 11:00AM, outlet moisture ranged
between -7.1% & -5.4% (since the moisture sensor was not calibrated
accurately at this time, negative readings were confirmed to be 0% grain),
and grain temperature ranged between 220 & 200 degrees Fahrenheit. At
11:00AM the set point was changed to 2800lbs/hr in order to cause a
change in outlet moisture. At 1:00PM and 4:00PM, moisture readings were
-7.8% & -1.6% respectively. According to one of the operators, sometime
between 6:30PM & 11:00PM, one of the Zenith Presses turned up (probably
through vibration) to 14 RPM resulting in an increase in feed rate into the
dryer. At approximately the same time, the reflux dropout on the north end of
the reflux auger plugged up resulting in wetter grain entering the dryer. The
absence of reflux was present for an undetermined amount of time. These
two significant disturbances resulted in wet grain exiting the dryer at
approximately 10:00PM forcing the operator to shut off the Zenith Presses.
Between 7:00PM & 1:00AM, the outlet moisture content ranged between
1.1% & 12.3%.
The following conclusions were made after analyzing the data
compiled during this test. Since the dryer ran a significant amount of time
early in the day with fairly wet inlet grain and a steam flow of 3000lbs/hr, and
did not show an increase in outlet moisture, it is quite possible that
3000lbs/hr or less is sufficient steam flow to dry the typical 40% to 50%
49


moisture grain instead of the 4000-4500lbs/hr steam flow currently being
used when the valve is wide open. Also, from the chart recorder and the
operators comments, it seems that the automatic control was working
properly until the two large disturbances took place simultaneously.
Consequently, even though the valve opened all the way, the dryer could
not heat up fast enough to dry the extremely wet grain entering at the
accelerated feed rate.
50


Date/ Time Inlet Moist. (%) Outlet Moist. (%) Steam Flow (Ibs/hr) Steam Press. (psi) Cond. Press. (psi) Grain Temp. (F) Stack Temp. (C) Zenith Press Speed (rpm)
3-1-94
8:00 76.7 1.1 4400 55 50 204 80 10
Set Point Set To 1800 lbs
8:15 77.0 1.1 1855 37 33 201 78 10
8:30 77.5 1.2 1803 22 18 180 76 10
8:45 77.8 1.3 1816 18 13 170 76 10
9:00 77.8 0.9 1807 17 12 168 76 10
9:15 77.6 1.6 1741 15 10 154 76 10
9:30 77.4 3.4 1807 15 11 144 76 10
9:45 77.4 4.0 1825 16 11 136 76 10
10:00 77.2 4.2 1807 17 13 140 76 10
10:15 77.0 5.5 1812 17 13 130 76 10
10:30 77.1 6.9 1798 17 13 130 76 10
10:45 77.0 7.3 1798 18 14 122 76 10
11:00 77.1 7.3 1772 18 14 134 76 10
Table 7.1. Spent Grain Dryer #4 Test Data Obtained on 3-1-94
51


Date/ Time Inlet Moist. (%) Outlet Moist. (%) Steam Flow (Ibs/hr) Steam Press. (psi) Cond. Press. (psi) Grain Temp. (F) Stack Temp. (C) Zenith Press Speed (rpm)
3-17-94
8:00 40.85 0.20 4100 55 48 220 80 10
Set Point Set To 2735 lbs
8:15 _ 0.02 2744 37 31 210 78 10
9:00 47.15 0.02 2740 29 25 180 78 10
9:15 _ 0.03 2718 28 24 190 78 10
9:30 _ 0.05 2722 27 23 180 78 10
9:45 _ 0.04 2731 26 22 180 78 10
10:00 47.35 0.07 2740 26 22 180 78 10
10:15 0.04 2735 26 22 170 78 .10
10:30 _ 1.00 2732 26 22 160 78 10
11:00 45.88 1.30 2732 26 22 160 78 10
11:15 1.90 2735 26 22 158 78 10
11:30 _ 2.60 2730 26 22 156 78 10
11:45 - 2.80 2863 26 22 156 78 10
Table 7.2. Spent Grain Dryer #4 Test Data Obtained on 3-17-94
52


Date/ Time Inlet Moist. (%) Outlet Moist. (%) Steam Flow (Ibs/hr) Steam Press. (psi) Cond. Press. (psi) Grain Temp. (F) Stack Temp. (C) Zenith Press Speed (rpm)
3-22-94
8:00 0.80 4400 56 51 220 80 10
Set Point Set To 2200 lbs
8:30 0.80 2200 29 24 180 78 10
9:00 _ 2.20 2207 25 20 170 78 10
Set Point Set To 2948 lbs
9:30 3.30 2946 29 22 160 78 10
9:45 _ 2.60 2942 30 25 169 78 10
10:00 _ 2.30 2946 31 26 169 78 10
10:15 _ 2.90 2937 31 26 169 78 10
10:30 _ 2.60 2955 30 25 . 179 78 10
10:45 _ 2.30 2950 29 23 175 78 10
11:00 _ 3.80 2942 28 22 160 78 10
11:15 _ 5.80 2937 26 21 . 146 78 10
11:30 - 6.00 2951 26 21 140 78 10
Table 7.3. Spent Grain Dryer #4 Test Data Obtained on 3-22-94
53


Date/ Time Inlet Moist. (%) Outlet Moist. (%) Steam Flow (Ibs/hr) Steam Press. (psi) Cond. Press. (psi) Grain Temp. (F) Stack Temp. (C) Zenith Press Speed (rpm)
3-23-94
8:00 _ 0.60 4000 56 51 220 80 10
Set Point Set To 2000 lbs
8:30 _ 0.60 2000 25 21 210 75 10
9:00 _ 0.80 2000 23 19 194 75 10
9:30 _ 1.40 2000 22 18 182 75 10
Set Point Set To 2667 lbs
10:00 _ 2.30 2664 30 26 190 80 10
10:30 _ 3.20 2673 30 26 160 80 10
11:00 5.00 2660 30 26 150 80 10
Table 7.4. Spent Grain Dryer #4 Test Data Obtained on 3-23-94
54


Date/ Time Inlet Moist. (%) Outlet Moist. (%) Steam Flow (Ibs/hr) Steam Press. (psi) Cond. Press. (psi) Grain Temp. (F) Stack Temp. (C) Zenith Press Speed (rpm)
5-20-94
8:00 _ 0.10 3600 58 52 240 80 10
Set Point Set To 2700 lbs
9:00 0.10 2704 37 31 220 78 10
10:00 0.40 2700 32 27 200 78 10
11:00 0.50 2709 34 29 194 78 10
1:00 _ 1.00 2696 29 25 182 78 10
2:00 1.90 . 2700 29 25 182 78 10
2:30 - 2.20 2700 29 25 170 78 10
Table 7.5. Spent Grain Dryer #4 Test Data Obtained on 5-20-94
55


Date/ Time Inlet Moist. (%) Outlet Moist. (%> Steam Flow (Ibs/hr) Steam Press. (psi) Cond. Press. (psi) Grain Temp. (F) Stack Temp. (C) Zenith Press Speed (rpm)
5-25-94
7:30 46.3 -0.1 2000 58 52 232 70 10
Set Point Set To 1800 lbs
9:00 44.0 -0.1 1494 39 33 220 75 10
9:30 . -0.1 1499 39 33 200 75 10
Set Point Set To 3000 lbs
10:00 40.4 0.5 2994 50 45 220 75 10
11:00 45.0 0.3 2981 58 52 240 80 10
1:00 46.8 -0.6 2753 60 55 240 80 10
3:00 -0.1 2665 60 55 240 80 10
5:00 -0.1 2901 60 54 240 80 10
5:30 _ 0.3 2545 60 54 240 80 10
7:00 0.2 _ _ _ _ -
9:00 - 1.9 - - - - -
Table 7.6. Spent Grain Dryer #4 Test Data Obtained on 5-25-94
56


Date/ Time Inlet Moist. (%) Outlet Moist. (%) Steam Flow (Ibs/hr) Steam Press. (psi) Cond. Press. (psi) Grain Temp. (F) Stack Temp. (C) Zenith Press Speed (rpm)
6-1-94
8:00 69.0 -7.1 3047 57 52 220 74 10
Set Point Set To 3000 lbs
9:00 69.0 -5.4 3000 53 48 217 74 10
10:00 74.0 -6.1 3000 50 45 217 74 10
11:00 71.0 -6.9 3000 47 42 200 74 10
Set Point Set To 2800 lbs
1:00 69.0 -7.8 2819 44 38 165 74 10
4:00 _ -1.6 2810 39 33 _ 74 10
7:00 _ 1.1 _ 38 _ 77 10
9:00 _ 1.4 _ 38 _ _ 76 14
11:00 7.2 - 56 _' _ 83 14
1:00 _ 12.3 - 56 _ 83 14
3:00 0.6 49 80 14
5:00 _ 1.3 _ 43 _ _ 80 14
7:00 0.3 - 44 _ _ 79 10
8:00 - 0.1 2800 45 40 - 79 10
Table 7.7. Spent Grain Dryer #4 Test Data Obtained on 6-1-94
57


8. Economic Considerations
Naturally, part of the intent of this research was to prove the feasibility
of a more efficient method of drying spent grain. To prove its feasibility,
dollar figures were required to illustrate the true savings. Following are
three different scenarios that illustrate the possible yearly savings of three
different types of control systems.
8.1 Proposed Control System #1:
0% Grain with Reduced Steam Usage
As stated earlier in this manuscript, grain is currently being dried to
0% by using the maximum steam available. For example, opening the valve
100% anytime the dryer is running. Through my experiments, I believe that
0% grain can be produced using significantly less steam. An indication of
this is that the 0% grain currently being produced has a surface temperature
of approximately 220 degrees Fahrenheit whereas the 0% grain exiting the
dryer at a reduced steam flow has a surface temperature of approximately
160degrees Fahrenheit. This illustrates that the extra steam is being used
primarily to increase the temperature of the grain but is doing nothing for the
drying or quality of the grain. In fact, this heating action contributes directly
to the increased fire hazard.
The first proposed system would require an outlet moisture sensor, a
steam flow transmitter, and a manually actuated Cam Flex steam valve. This
system would require that the operator manually adjust the steam valve
58


position to obtain approximately 3000lbs/hr of steam flow. Following is a
cost estimate of the potential savings of installing such a control system that
would require the minimal amount of equipment, and therefore would be the
least expensive system to implement.
Current Steam Usage = 4500 Ibs/hr
Required Steam Usage = 3000 Ibs/hr
Steam Usage Difference = 1500 Ibs/hr
Operating 24 hrs/day, 365 days/yr, and with steam costing $3.50/1000 lbs
translates into a savings of:
lbs hrs days
1500 TTr~ 24 dly' 365 year
$3.50
10OOIbs
$45,990.00/yr
There would not be a significant savings in operator time because operators
would still be required to take hourly readings of steam pressure,
condensate pressure, and stack temperature.
8.2 Proposed Control System #2:
8% or Less Grain
The second proposed control system operates by restricting the outlet
grain moisture to 8% or less. Naturally, part of the time the grain will be over
dried, but considering the current nonexistence of a control system, this
alternative definitely is an improvement. This system would require a DCS,
an outlet moisture sensor, a Cam Flex steam valve, and an alarm system to
alert operators of a problem. This system can be implemented utilizing
59


feedback techniques exclusively. Basically, the steam valve position would
be set at a high enough setting to dry incoming grain regardless of inlet
conditions. Results of my tests indicate that a 75% open valve provides
enough steam to dry the grain to less than 8% regularly. The outlet moisture
sensor would be used as an interlock alarm device in that if a reading of 8%
grain resulted on the sensor, the controller would open the valve 100% until
the 8% criteria was met. The following cost estimate illustrates the potential
savings of this type of system:
Steam Used With a 100% Open Valve = 4500 Ibs/hr
Steam Used With a 75% Open Valve = 2500 Ibs/hr
Difference In Steam Usage = 2000 Ibs/hr
Operating 24 hrs/day, 365 days/yr, and with steam costing $3.50/1000 lbs to
produce, translates into a savings of:
Note that this figure does not the reflect the savings of the decreased steam
usage in producing pellets.
8.3 Proposed Control System #3:
The final proposed system would return the biggest savings but would
also be the most expensive to install. It involves encompassing the entire
control system researched and the maximum number of data acquisition
measurement devices available. It involves the use of feedback control,
days $3.50
'JOS * -tnnmu
year 1000lbs
= $61,320.00/yr
6%-10% Grain
60


feedforward control, adaptive control, self-tuning control, and model
reference control. This system would require a DCS, an inlet and outlet
moisture sensor, an inlet grain flow meter, a Cam Flex steam valve, a steam
flow transmitter, and an alarm system to alert operators of a problem. The
\
individual stages of this system were tested repeatedly, independently of
each other and near the conclusion of this research, multiple stages were
being tested together. This system would have a target outlet moisture of
8% and could realistically keep the outlet grain moisture between 6% and
10%. The feedforward information would be used to detect changing inlet
conditions and the feedback information would be used both to control the
system, but also to override the system in an emergency situation i.e. rapidly
rising outlet grain moisture. Taking the steam savings estimation of
$61,320.00/yr from Control System #2 and combining that with the solids
giveaway of the pellets that is calculated below:
Outlet Feed Rate = 4000 Ibs/hr
Actual Outlet Moisture = 1%
Target Outlet Moisture = 8%
Moisture Leaving Dryer @ 8% = 320 Ibs/hr
Moisture Leaving Dryer @ 1% = 40 Ibs/hr
Difference (Solids Giveway) = 280 Ibs/hr
Operating 24 hrs/day, 365 days/yr, and with pellets selling for an average of
$90.00/ton translates into a lost revenue of:
61


panics P4hrs. days 1 ton $90.00
^au hr day afc,:3 year' 2000lbs ton
$110,376.00/yr
The total yearly savings of this system would be:
Steam Savings = $61,320.00/yr
Solids Giveaway Savings = $110.376.00/vr
$171,696.00/yr
Therefore, any one of the proposed control system would be a
significant improvement over the manual system currently being used.
62


9. Conclusions
On the basis of this research including the tests conducted, I feel that
there is great potential for implementing any one of the suggested control
systems. Also, like many other companies, Adolph Coors Company is
committed to identifying large inefficiencies throughout its processes, and
the spent grain dryer is clearly one of these areas. Based on the dollar
figures to implement an automatic control system, the project should be
seriously considered for installation.
Further research is warranted in this area because it seems to be a
common problem in many food processing or grain processing facilities and
therefore would be a great benefit to not only Adolph Coors Company, but
also to any facility that requires better control of their material drying process.
63


REFERENCES
(1) Myron, T.J. and Shinskey, F.G., "Product Moisture Control for Steam-
Tube & Direct Fired Dryers", MBAA Technical Quarterly, Vol. 12, No. 4,
pp. 235-242, 1975.
(2) Isdale, C.E., "Opportunities for Process Control in the Cereal
Industries", American Association of Cereal Chemists, Inc., Vol. 36,
No. 4, pp. 364-367, April 1991.
(3) McFarlane, N.J.B., "Control of High-temperature Grain Dryers",
Measurement and Control, Vol. 26, pp. 206-209, September 1993.
(4) Endress and Hauser, Granumet DE10 and DME270 Impact Weigher
Installation and Operation Manual, No. IM 003G/01/ae/04,93.
(5) Sensortech Systems Inc., ST-2200A Technical Manual.
64