Beruflich Dokumente
Kultur Dokumente
Abstract—New experimental procedures are being developed learn concepts of incompressible viscous flow, including
to help students connect course theory to real-world systems. energy considerations and head loss calculations. Faculty and
The time constant of a fluid system consisting of a tank, inlet undergraduate research assistants are redesigning the
flow, and outlet flow through a valve is determined by the Feedback Control Systems course to include a significant
resistance and tank area. The valve resistance is non-linear and
laboratory component.
depends on the flow rate and tank level. Both the time constant
around an operational point and the average time constant This project is part of a larger initiative that seeks to offer
through a range of operational points can be determined through students a more complete understanding of mechanical
experimental and analytical techniques. The time constant of an
engineering concepts by providing hands-on experiences and
experimental liquid-level rig is determined by deriving a linear
model of the system and experimentally determining the
showing connections to professional practice [1]. Inductive
relationship between steady-state outlet flow rate and tank level. learning techniques, which engage students through hands-on
These analytical and experimental methods incorporate concepts experimental procedures and practical applications are being
mechanical engineering students learn in several core courses. introduced into the Mechanical Engineering Department’s
Homework assignments and laboratory experiments exploring a Feedback Control Systems Course. Since 2008, the class has
physical system serve to engage students and help them make incorporated a significant laboratory portion [2]. In these labs,
connections between course material spanning the mechanical students become familiar with and perform experiments on
engineering curriculum. two educational rigs from Feedback, Inc. The first rig, the 38-
600 Thermal Process Rig, is a thermal system in which the
Keywords—laboratory design; time constant; fluid dynamics;
control systems; linearization; valve resistance temperature of the inlet and outlet flow can be controlled by
varying the flow rate through a heat exchanger. The second
I. INTRODUCTION rig, the 38-100 Basic Process Rig, allows students to control
This project explores several methods of determining the flow and level throughout a fluid system. This project serves
system time constant for a dynamic fluid system through both to improve upon the current set up of this lab by more closely
analytical and experimental techniques. The system, which associating the experiments with material presented in lecture
includes a tank, an inlet flow, and an outlet flow through a and in other classes, including Systems Engineering and Fluid
valve of unknown resistance, is modeled using approaches Mechanics.
taught in the Systems Engineering, Feedback Control Systems II. SYSTEM OVERVIEW
and Fluid Mechanics courses. Systems Engineering is a
sophomore course which introduces the mathematical The system under consideration for this project is the
modeling and time response analysis of mechanical, Feedback Basic Process Rig 38-100 [3]. The set up, shown in
electromechanical, fluid, and thermal systems. Feedback Fig. 1, includes the right side of the tank, a solenoid valve, a
Control Systems is taken junior year and is a continuation of servo valve, a centrifugal pump, a visual flow meter, and
Systems Engineering. In Feedback Control Systems students copper piping. The right side of the tank has a cross-sectional
learn various methods of analyzing and controlling linear area of 205 cm2. An empty tank corresponds to 0% level as
time-invariant dynamic systems. Students take Fluid designated by the ruler on the left side. The tank at 100%
Mechanics concurrently with Feedback Control Systems and level is 22.25 cm above the outlet valve. The solenoid valve,
with an inner diameter of 6 mm, is open for all experiments.
Manuscript received February 28, 2014. This research is supported by An 8.9 cm copper pipe runs between the bottom of the tank
NSF TUES DUE No. 1044830.
and the solenoid valve. The position of the gate within the
A. Zielkowski is a Mechanical Engineering student and a research
assistant helping to develop new undergraduate laboratory experiences at The servo vale is controlled by sending a varying current signal
Cooper Union for the Advancement of Science and Art, New York, NY between 4 mA and 20 mA, where 4 mA corresponds to the
10003 USA (email: zielko@cooper.edu). valve being 0% open, and 20 mA corresponds to the valve
M. Baglione and D. Wootton are with the Mechanical Engineering
Department, The Cooper Union for the Advancement of Science and Art,
New York, NY 10003 USA (email: melody@cooper.edu or
wootto@cooper.edu).
being 100% open. The servo valve controls the inlet flow the head and flow rate increase, so does the resistance of the
rate, which is driven through the cooper pipes by the outlet valve. The slope of the graph is smaller in the laminar
centrifugal pump. The visual flow meter, which measures the flow range and increases greatly as the system becomes
inlet flow rate, is able to read flow rates between about 1.0 turbulent.
L/min and 4.0 L/min in 0.2 L/min increments.
Fig 1: (a) Liquid-Level System; (b) head versus flow rate curve [2]
𝑑ℎ 1
𝐴 + ℎ = 𝑞!" (1)
𝑑𝑡 𝑅
Fig. 3: Head versus flow rate. Fitting the experimental data using linear
where A is the cross-sectional area of the tank, R is the regression trendlines and finding the slopes yields the resistance with a y-
resistance of the outlet valve, and h is the head, or level of intercept=0 (ya) and resistance for flow rates between 1.25 and 1.7 L/min , or
𝐜𝐦 𝟑 𝐜𝐦 𝟑
fluid in the tank, the time constant, 𝜏, can be calculated by the roughly 20
𝐬
to 27
𝐬
(yb).
following:
it is apparent that
𝐿 16 𝑞!"
𝑅 = 𝐾! + 1 + 𝑓
(26)
𝐷!"!# 𝜋 𝐷!"!# ! 𝑔
!
𝑐𝑚!
8.9𝑐𝑚 16 (23.7)
𝑅 = 81.86 + 1 + (0.046) 𝑠
1.26𝑐𝑚 ! ! 𝑚
𝜋 12.6𝑚𝑚 (9.81 ! )
𝑠
𝑠 𝑠
𝑅 = 13180 = 1.32
𝑚! 𝑐𝑚!
The minor losses in the system are due to the flow through E. Method 5: Initial condition with inlet flow rate yielding
the outlet valve and can be determined by the following steady-state level of 50%
equation: Methods 2, 3 and 4 look at operating points around higher
𝐾! 𝑉 ! flow rates, where the resistance is roughly linear. Method 1
𝐻!"#$% = (19) looks at the operating range of the entire system, which is
2𝑔
overall highly nonlinear. Due to the roughly linear behavior
where 𝐾! is a constant related to the resistance of the outlet in this system across higher flow rates, as show in Fig. 2(b), a
valve. The velocity at point 1 is 0 cm/s. The velocity at point time constant determined experimentally for operational
! !
2 is !"# , which is equal to !" , at steady-state. Noting 𝑝! = points across higher flow rates only is expected to more
!!"!# !!"!#
closely match a time constant determined analytically by
𝑝! = 𝑝!"# , and ℎ! = 0, Bernoulli’s equation simplifies to:
assuming linearity in this range, such as in Methods 2 and 3.
𝑉! ! An experiment similar to Method 1 was run, looking only at
ℎ! − = 𝐻! (20)
2𝑔 the operating range between 100% to 50% level. Similar to
Method 1, the tank was filled to 100% level as an initial
The entire energy balance for the system is:
𝐿 condition. Instead of observing the zero input response, an
𝑓 𝑉! !"!
𝑉! ! 𝐷!"!# ! 𝐾! 𝑉 ! (21) input flow rate of 1.5 L/m (23.8 ! ), for which the steady-
ℎ! − = +
state value is 50% level, was supplied to the tank. To measure
2𝑔 2𝑔 2𝑔
Solving for 𝐾! yields: the time constant, the time for the tank to drain from 100%
ℎ! 2𝑔 𝐿 level to 66% level, or 63% to steady-state, was timed. By this
𝐾! = − 1 + 𝑓
(22) method, τ was determined to be 230 seconds.
𝑉 ! 𝐷!"!#
IV. RESULTS AND DISCUSSION
ℎ! 𝑔𝜋 ! 𝐷!"!# ! 𝐿 The time constants determined by each method are listed in
𝐾! = − 1 + 𝑓
𝑞!" ! 8 𝐷!"!# (23) Table I. It is interesting that both linearization, a technique
taught in Systems Engineering, and an energy loss
Using the same operational point that was considered in consideration, a method taught in Fluid Mechanics, yielded
Method 2, similar results around a single operational point. The time
!
!".!!" !.!" !! (!".!!!)! !.!!" constant found experimentally in Method 5 varies from these
!!
𝐾! = ! !
− 1 + (0.046)
!.!"! ! !.!" !" time constants, but is still similar. This validates both
linearization and the energy loss consideration, as both
𝐾! = 81.9 methods are reasonable approximations of the system.
Methods 1 and 2 yielded far different results for the system Engineering asks students to consider linearization of a system
time constants, most prominently because these methods are around an operational point to model system responses. Fluid
looking at a larger range of operational points. As Fig. 2(b) Mechanics gives students tools to consider energy losses in
shows, R changes drastically depending on the flow rate and pipe flows. Feedback Control Systems offers students hands-
level the system is being operated around. Method 1 on experience with physical systems so that they may compare
considers all values of R, as it requires the system to go from modeled behavior with actual behavior.
100% level to 0% level. Theoretically, the value of τ from The current proposal for this project asks students to
Method 2A, using the slope of the trendline which goes complete Method 3 as a homework assignment in Systems
through the origin, should yield a similar value as Method 1. Engineering class, given the experimental data for the
However, due to the measurement deadband of the visual flow operational point at a flow rate of 1.45 L/m and the diagram of
meter at low input flow rates, there is data missing from the the system shown in Fig. 4. The following semester, in Fluid
system representation in Fig. 3. If this data could be collected, Mechanics, the students will derive an additional model of the
it would likely change the linear trendline calculated by system, following Method 4, again looking at the operation
Microsoft Excel, which would in turn change the resistance point at a flow rate of 1.45 L/m. Concurrently, in Feedback
value and time constant calculated by this method. Still, there Control Systems, students will experimentally determine τ,
would be a discrepancy between the values, as Method 2A using Methods 1 and 2 using the 38-100 Basic Process Rig.
assumes a linear system, which the system is obviously not. They will be asked to think about why there are discrepancies
between the two values and hopefully realize the
TABLE I consequences of modeling a nonlinear system as linear. After
EXPERIMENTAL TIME CONSTANT RESULTS AND CALCULATIONS comparing the results, students will use the experimental
Time constant,
Method values of τ to choose PID control gains that achieve the
τ (seconds)
1: Initial tank level free response 57 desired liquid-level response. In this way, students validate
2A: Determining resistance from head versus flow the accuracy of their models with experimentally determined
133
rate relationship (y-intercept=0) values of the time constant, τ, and implement their findings in
2B: Determining resistance from head versus flow a real-time control system. Synthesizing concepts from
334
rate relationship (average turbulent region)
different mechanical engineering courses and allowing
3: Linearizing around a single operational point 265
4: Utilizing energy considerations in pipe flow 270 students the opportunity to experiment with and validate these
5: Initial condition with inlet flow rate yielding concepts aims to provide students with a more concrete
230
steady-state level of 50% understanding of the theory they learn.