Anda di halaman 1dari 10

Development of a Stepless Flow

Control System
By
R. Aigner, A. Voser, A. Allenspach
Burckhardt Compression AG
Winterthur
Switzerland
roland.aigner@burckhardtcompression.com

7th Conference of the EFRC


October 21th / 22th, 2010, Florence

Abstract:
Capacity control systems are used to control the amount of gas delivered by a compressor and/or to
ensure certain pressure levels at the intermediate stages or at the outlet of the compressor. In the
present study, a new approach for a capacity control will be presented and compared to existing
systems. It turns out that the new approach is superior in major aspects, such as available flow rate,
amount of energy saved and reliability. The benefits are verified with simulations showing the impact
of the flow control on valve dynamics and fluid dynamics and with measurements on compressors in
operation.

Introduction

The operation of large reciprocating compressors


often requires energy- and cost-efficient ways to
adjust mass flow and/or to ensure certain pressure
levels at the intermediate stages or at the outlet of
the compressor. A wide variety of approaches are
available. The most prominent are:

Bypass: allowing the already compressed gas


to expand to the suction side of a compressor
stage through a dedicated gas pipe equipped
with a bypass valve.

Clearance pocket: compressing gas into


additional clearance pockets attached to a
compression chamber.

Speed control: varying the speed of the


compressor.

Suction valve lifting: keeping the suction


valve open with a so-called unloader.

A detailed description, advantages and drawbacks


can be found in reference 11 and only the latter
method will be discussed further. Here, the already
suctioned gas is delivered back to the suction line
and this can be performed either during one or
more subsequent full strokes as intermittent flow
control or only for some portion of the
compression phase as so-called stepless reverse
flow control. According to Figure 1, which
compares the efficiency of the capacity control
methods described above, the intermittent flow
control is more efficient than stepless flow control.
On the other hand, stepless reverse flow control, as
the name suggests, can adjust the mass flow and
the pressure levels continuously variably.
The capacity control presented in this paper allows
intermittent and stepless flow control to be
combined in order to maximize efficiency and
reliability. The unloader is driven by a magnetic
actuator, similar to devices already presented by
Kopecek, Klockow, Schmitz (2008) 2 and
Schiavone, Ragi (2008) 3. The simple and therefore
cost-effective design and the availability of rapid
design tools for the simulation and control of such
mechatronic devices make this the preferred choice.
In order to ensure maximum reliability, a complete
simulation of the capacity control system has been
used for the development of the unloader device.
Furthermore, long-term test runs of prototypes have
been conducted on different test compressors in
operation.

Figure 1: Power consumption of the compressor


depending on the mass flow and the flow control
method used.

Description of the stepless capacity


control system

2.1 Configuration of the stepless


capacity control system
The main components are as follows:

Unloader

Magnetic actuator

Position sensor

Power electronics and control unit

Master control unit and power supply

The compression chamber is separated from the


suction side of the gas chamber by a conventional
passive suction valve (Figure 2). The fingers of the
unloader reach the sealing elements of the valve
through the valve seat and can push them against
the valve guard (open valve). In Figure 2, a plate
valve is shown, but any kind of valve can be used
(plate valve, poppet valve, ring valve). The
unloader itself can be pushed down by the plunger
of the magnetic actuator.
The magnetic actuator is deactivated in the initial
position, and the spring holds the unloader and the
armature of the magnet in the uppermost position.
Therefore, the valve works without any influences.
The armature is pulled down if an electric current is
running through the coil of the magnet and the
plunger pushes the unloader down. The unloader,
in turn, pushes against the valve plate, and
therefore, the valve can be kept open, even if the

2/10

pressure force of the gas flow tries to close the


valve.
In order to control the motion of the unloader, a
position sensor determines the state of the armature.
This signal is fed through a control unit to
determine the desired electric current of the
magnetic coil. In addition a master control unit
coordinates the necessary control parameters and
information (e.g.: crank angle) for the control units
at each suction valve.

reaches the bottom dead centre (BDC). In the case


of activated stepless reverse control, the suction
valve is kept open beyond the beginning of the
compression phase by means of the unloader.
Therefore, a part of the fluid sucked into the
compression chamber is pressed back into the gas
chamber of the suction side. At a certain crank
angle, the magnetic actuator is deactivated and the
unloader moves towards the initial position. Due to
the flow forces, the valve sealing element (e.g.
valve plate) is pressed onto the unloader, causing it
to close. The flow forces can be much higher than
the spring forces, especially with heavy gases, high
suction pressures and high-speed compressors. The
unloader is used to control the motion of the sealing
element. In this example, the sealing element is put
gently on the valve seat at a crank angle of 290.
Again, the impact speed must be small to achieve
soft landing. The time of closing the suction valve
determines the quantity of the fluid remaining in
the compression chamber, and thus the mass flow
of the gas delivered. Finally, the unloader returns to
its initial position.
Figure 5 shows the thermodynamic cycle in the
pressure-volume diagram. The red line corresponds
to the reverse flow control. The area enclosed by
this line represents the thermodynamic work of the
process. The difference in the areas of the full-load
cycle (green line) and the reverse flow control
cycle is equal to the amount of saved energy.

Figure 2: Main components of the stepless capacity


control system

2.2
2.2.1

Operation of the capacity control


system
Stepless reverse flow control

The unloader is in the initial position during the


expansion phase of the compression chamber and
the valve is closed (Figure 3). The resulting
pressure force acting on the valve sealing element
is the pressure difference between the compression
chamber and the suction side multiplied by the
effective force area. The valve starts to open when
the resulting pressure force on the valve sealing is
higher than the spring force (crank angle, CA=90).
Then the magnetic actuator is activated and the
unloader moves in the direction of the valve guard.
At a crank angle of approximately 145, the
unloader reaches the sealing element and presses it
against the valve guard. Here the impact velocity
must be smaller than 0.1 m/s (see section 2.2.4 Soft
landing).
In the case of uncontrolled valves, the suction valve
would start to close at approximately 150 CA. The
valve would be firmly closed when the piston

Figure 3: Reverse flow control: motion of the


unloader and valve plate during one compression
cycle.

2.2.2

Intermittent capacity control

In the case of intermittent flow control, the suction


valve is kept open for a whole cycle (Figure 4). All
the gas sucked into the compression chamber is
pressed back to the suction side during the
compression phase. This compression chamber
then delivers no gas in this cycle. In order to

3/10

deliver approximately 50% of the mass, the suction


valve is kept open every other cycle. Almost any
mass flow and/or intermediate pressure can be set
by choosing the number of working and idle cycles.
Due to the pressure losses of the gas flowing in and
out of the compression chamber, a small amount of
energy is spent. The corresponding thermodynamic
work can be found in Figure 5 (area enclosed by
the blue line).

Calculations and experiences have shown that the


discharge valves have shorter life times for this
load. There are two reasons: firstly, the discharge
valve does not work at the designed operating
point, which can lead to late closings of the valve,
unpredictable movements and loads of the sealing
element. Secondly, the closing impact velocities of
the valve sealing element can increase (Figures 6
and 7).
Intermittent capacity control is slightly more
efficient (Figure 1). Furthermore, the durability of
the control components increases since not every
cycle is controlled. The drawback is that the mass
flow can only be adjusted gradually.
A combined capacity control functions as follows:
High mass flows: Basically, stepless reverse
flow control is used. But in this case, the
magnetic actuator is not activated every cycle,
so that the suction valve can operate one or two
cycles without influences. The advantage is that
the durability of the components increases and
the capacity control itself uses less energy.

Figure 4: Intermittent capacity control: motion of


the unloader and valve plate during one
compression cycle.

Medium mass flows: The reverse flow control


is activated in every cycle.
Low mass flows: Basically the intermittent
capacity control is used and the suction valve is
kept open for one or two cycles. Moreover, the
retreating of the unloader can be delayed
(reverse flow control) and stepless adjustment
of the compressor capacity is possible.

Figure 5: Pressure-volume diagram: different


capacity controls

2.2.3

Combined capacity control:


Intermittent control and reverse
flow control

Combining the intermittent control and stepless


reverse flow control also combines their advantages
and their disadvantages vanish.

Figure 6: Valve plate lift: 100% mass flow when


the capacity control is switched off. 18% mass flow
when the reverse flow control is used.

The main advantage of reverse flow control is that


a continuously variable mass flow can be set.
However, small capacities cause difficulties.

4/10

Figure 7: Influence of the reverse flow control on


the discharge valve. *Impact velocity related to
allowed impact velocity.

2.2.4

Soft landing

The term soft landing refers to a motion where


the impact velocities of a moving part (the valve
sealing element or the unloader) are below 0.1 m/s.
Here, the limit of 0.1 m/s has been chosen because
noise and wear are considered to be low. In other
words, soft landing guarantees high durability. In
addition, in cases where low noise levels are
required, the capacity control does not add much
noise (Figure 8).

Figure 8: Measurement of the sound level (distance


to capacity control: 0.5 m) depending on the
impact velocity without compressor noises.
High impact velocities lead to wear within short
time periods (Figure 9). The impact of the unloader
on the valve sealing element is especially critical.
The selected contact materials also play an
important role.

2.3
2.3.1

Components of the capacity


control
Magnetic actuator

A wide range of actuators are available to operate


the unloader. Not only magnetic actuators but also
hydraulic actuators provide the necessary main
features.

Figure 9: Wear of the valve sealing element caused


by the unloader after 200 hours of operation with
high impact velocities.
The advantage of the hydraulic actuator is its
compact design. However, the drawbacks outweigh
the advantages of this solution. Firstly, the
hydraulic system has high acquisition and
maintenance costs. Secondly, it is prone to oil
leakage. On the other hand, the advantage of the
electromagnetic actuator is that rapid design
changes of the motion control are possible and
rapid design tools for simulation and control are
available. However, soft landing requires advanced
control approaches. Everything considered, a
magnetic actuator was chosen.
The development of an electromagnetic actuator is
based on the most-used concept of electromagnetic
controlled valves in the automotive industry 4, 5, 6.
In Figure 10 / Magnet, A the basic principle of such
an actuator is shown. Two springs force the
armature of the magnet to be in a middle position.
A small retaining magnet is placed at each side of
the armature. A typical cycle proceeds as follows:
The armature is held in an end position by the
retaining magnet. When the magnet is switched off,
the springs accelerate the armature towards the
opposite end position. The armature runs through
the central position at maximum speed and is then
slowed down because of the springs. Finally, when
the armature reaches the opposite end position, the
second retaining magnet is switched on and the
armature is captured and held. When using this
system in reciprocating compressors, three
problems occur:

When using reverse flow control, the actuator


should slow down the movement of the valve
sealing element and should put the sealing
element gently onto the valve seat. Since this
actuator cannot stop the armature between the
two end positions, the sealing element of the
valve has a high impact velocity when the
valve closes.

5/10

As the valve closes, the flow force can be


much higher than the friction and the spring
forces. Hence the armature is accelerated
beyond the middle position, resulting in high
armature impact velocities. Soft landing is not
possible since the system is not capable of
decelerating the movement.

If the armature does not reach the opposite


end position due to increased friction or a
malfunctioning motion control system, the
armature ends up in the middle position and
leaves the valve half open, which may cause
problems. Furthermore, the initialization of
the magnet (moving the armature back to an
end position again) will take some time.

In order to overcome the problems stated above, the


two magnets have been reduced to one stronger
magnet, which is capable of applying enough force
on the armature over the whole lift (Figure 10,
Magnet B). In addition, only one spring is
necessary to provide the returning force.

Taking everything into consideration, Magnet C


was chosen.

Figure 11: Final solenoid configuration: Magnet C


has one working air gap in the magnetic circuit.

2.3.2 Unloader
There are two main technical requirements for the
unloader (Figure 12). Firstly, the unloader must
have low weight in order to reduce the cycle time
of the capacity control and, in this context, the
required force of the actuator. Secondly, a low drag
coefficient reduces the required force of the
actuator, guarantees unaffected operation of the
valve during idle times, and keeps the pressure
losses, and therefore the energy losses, low.

Figure 10: Different armature configurations:


Magnet A consists of two magnetic actuators. The
springs hold the armature in centered position.
Magnet B consists of one magnet only and the
returning force is derived from the spring mounted
on the unloader.
Two different solenoid configurations have been
considered (Figure 10, Magnet B and Figure 11,
Magnet C). Only one working air gap is in the
magnetic circuit of Magnet C. This leads to
increased forces at high lifts and slightly decreased
forces at low lifts compared to Magnet B. The
drawback of design C is the higher load of the
bearings due to the transverse forces in case of
imperfect symmetry. In comparison, Magnet B has
two working air gaps, which lead to small
transverse forces. However the magnet has to be
bigger to achieve the same force at high lifts.

Figure 12: Unloader for the Burckhardt poppet


valveTM.

6/10

2.3.3 Position sensor


The motion control quality strongly depends on the
performance of the position sensor of the magnetic
actuator. High standards must be demanded
regarding linearity, resolution, frequency response
and temperature stability, to name but a few. An
eddy current sensor or an incremental magnetic
position system fulfils these requirements.

Figure 13: Motion control: r=reference signal


(position), e=error, uF=output of feedforward
controller, uC=output of feedback controller,
u=total plant input (current), y=output (motion).

2.3.4 Power electronics and control unit


The power electronics and control unit determine
and deliver the required amount of electric current
for the magnet coil based on the actual position.
In order to achieve soft landing (see section 2.2.4)
the armature of the magnet must follow a given
position trajectory, and must therefore be provided
with the corresponding electric current. However, if
the electric current or the position of the armature is
changed (both effects alter the inductivity of the
magnet), eddy currents prevent the immediate
change of the magnetic force. In other words, the
build-up of magnetic force is delayed in
comparison to the current running through the
coils. Therefore, experiments have shown that a
simple PID-controller (proportional integral derivative controller) in closed-loop control
achieves only poor results. The armature cannot
follow the desired trajectory and the impact
velocities are high.
Performance improvements of the trajectory
following can be obtained with an additional
feedforward loop (Figure 13). With this, the
behaviour of the system can be altered in advance,
depending on the reference signal r. Robustness
against disturbances is warranted by the feedback
controller, which compensates the error e, the
difference between reference position r and the
actual position y. The total input u to the plant is
the sum of the feed-forward signal uF and the
feedback signal uC, representing the necessary
electric current. This signal is sent to a chopper that
delivers the current to the coils of the magnet 7.

2.3.5 Power supply and master control unit


The power supply and master control unit can be
located far away from the compressor (e.g.: outside
the explosive area). It supplies the necessary
control parameters and information (e.g. crank
angle) for the control units at each suction valve.
For diagnosis and monitoring purposes, the control
units send their status, error and monitoring signals
to the master control.

3
3.1

Simulation
Dynamic model

A dynamic model of the capacity control takes the


following subsystems into account:

Compression chamber and piping (fluid


dynamics, thermodynamics).

Valve dynamics.

Dynamics of unloader, spring and magnetic


actuator.

Controller
unit
(electrodynamics).

and

chopper

The modelling of the compression chamber, piping


and valve is well-known and a detailed description
can be found in the references 8,9,10. Moreover the
modelling of the unloader and spring dynamics,
electronics for the controller unit and chopper are
straightforward and will not be discussed further.
Three different approaches are used to capture the
behaviour of the magnetic actuator 1,7,.
Firstly, the governing equations of the magnetic
field are solved by means of finite element methods
(FE methods). Comparison of simulation and
measurement results shows that differences in static
magnetic forces are very small. Hence, this
approach is used to optimise the static
specifications of the magnet (size of magnet,
magnetic flux density of the ferromagnetic circuit
(Figure 14) and number of windings).
Secondly, an analytical function is used where the
static magnetic force depends on the armature lift
and electric current. This equation is extended with
terms describing the dynamic build-up of the
magnetic force. Even eddy currents can be taken
into account. Since the structure of the dynamic
model is known, model-based feedback and feedforward controllers can be designed and optimised.

7/10

this case, the maximum temperature of the coil does


not

Figure 14: Magnetostatic FE-simulation 2D:


magnetic flux density B of the magnet.
Thirdly, a lumped magnetic network is used. Here
the ferromagnetic circuit and the working air gap
are replaced by magnetic reluctances. Again, eddy
currents can be taken into account. Short
calculation times are the advantage of this method
compared to FE methods. However, differences
between calculated and measured forces can be
remarkable but sufficiently small for simulating the
complete system of the capacity control.
The results of the complete simulation of the
unloader lift, velocity and coil current show that
soft landing should be possible (Figure 15).
Observing the electric current trajectory gives
better insight into the magnetic actuator. High
electric currents are needed in order to set the
unloader in motion. During the motion towards the
valve guard, the current is reduced (90 CA). Since
the unloader lift decreases, the necessary current is
very low for holding the unloader in the end
position. In this case, the pressure forces on the
valve sealing element are rather high compared to
the spring forces. Therefore, a high amount of
electric current is needed to achieve softlanding
during the backward motion.

Figure 15: Simulation results: lift and velocity of


the unloader and current through the coil.
exceed 53C when the ambient temperature is
assumed to be 30C.

Forced convection has a much higher heat transfer


coefficient. Figure 16 shows a simplified model of
the capacity control, including a pressure vessel
where air is forced through cooling channels. In
this case, the maximum temperature of the coil does
not exceed 53C when the ambient temperature is
assumed to be 30C.

3.2 Thermal simulations


The electric current running through the coil
produces heat (up to approximately 200 W). CFD
(computational fluid dynamics) simulations have
shown that natural convection is not sufficient to
keep the temperatures of the capacity control
system below a critical temperature. Not only must
the coil and surface temperatures be limited, but the
power electronics and control unit also cannot
endure high temperatures.
Forced convection has a much higher heat transfer
coefficient. Figure 16 shows a simplified model of
the capacity control, including a pressure vessel
where air is forced through cooling channels. In

Figure 16: CFD simulation: cooling system of the


capacity control, forced convection.

Measurements

Long-term test runs have been conducted on two


different test compressors in operation to test the
functionality of the prototypes and verify the
simulation results. The first test was conducted on a
single-stage, double-acting compressor, where air is

8/10

compressed from 1 to 4 bar. Here, the design of the


capacity control and the motion control was
optimised. It turned out that using a slightly
different approach motion for the unloader
improves the impact velocity considerably. Instead
of moving the unloader towards the valve guard all
at once, the unloader comes to a brief halt before
reaching the final end position next to the valve
guard (Figure 17). In this case, the unloader devices
were fitted to plate valves with a lift of 1.35 mm.
Therefore, the lift of the magnetic actuator was
limited to 1.55 mm. The opening impact velocity is
kept below 0.1 m/s (0.45 m/s) and hence soft
landing is achieved. The closing impact velocity is
slightly higher (0.75 m/s, Figure 18)

Figure 19: Capacity control mounted on a twostage test compressor.

Summary and outlook

The development of a capacity control system was


presented in detail. Not only the basic principles of
such a device, but also the main design problems
were discussed.

Figure 17: Measurement of the unloader lift and


velocity during the motion towards the valve guard.

The investigation showed that an electromagnetic


system actuating unloaders has many advantages.
Furthermore, the best way of operating the
unloaders (regarding reliability, durability and
efficiency) is choosing between intermittent and
reverse flow control, depending on the desired
mass flow. The advantages of both operation
methods are used and the drawbacks vanish.
An advanced control unit is used to control the
motion of the unloader in order to achieve soft
landing. Soft landing, in turn, guarantees the
reliable and durable operation of the capacity
control.
The basic principles of the simulation of such a
system were discussed and important simulation
results highlighted. It turned out that forced
convection has to be used in order to limit the
operating temperature range of the magnetic
actuator and the power electronics.

Figure 18: Measurement of the unloader lift and


velocity during the motion away from the valve
guard.
In order to test the impact of the capacity control on
a multi-stage compressor, a two-stage compressor
was fitted with a set of prototypes (Figure 19).
Moreover, a long-term test run (500 hours) was
conducted and every component was analysed
afterwards. Not only the mechanical parts, but also
the used control were verified to function reliably.

The prototypes were tested and optimised on two


different compressors. The simulation results were
verified and the functionality of the mechanical
parts and the motion control were proven.
In the near future, field tests must be conducted and
approval for operation in explosive areas must be
obtained.

9/10

Acknowledgements

The authors want to thank P. de Lapersonne and M.


Lehmann for many fruitful discussions.

References
1

Kallenbach, E., Eick, R., Quendt P., Strhla, T.,


Feindt, K., Kallenbach, M. (2008): Elektromagnete.
Vieweg Teubner.
2

Kopecek, H., Klockow, H., Schmitz M. (2008):


Development and Test of an Electrical Valve
Actuator for Reverse Flow Capacity Control of
Reciprocating Compressors. 6th EFRC-Conference,
98-104.

Schiavone, M., Raggi, A. (2008):


Electromecahnical Actuator for Reciprocating
Compressor Stepless Control. 6th EFRCConference, 180-187.

Hoffmann W., Peterson, K., Stefanopoulou, A. G.


(2003): Iterative learning control for soft landing of
electromechanical valve actuator in camless
engines. IEEE trans on control system technology,
11(2), 174184.

Peterson, K., Stefanopoulou, A. G. (2004):


Extremum seeking control for soft landing of an
electromechanical valve actuator. Automatica, vol.
40, no. 6, pp. 10631069.

Tsai, J., Koch, Ch. R., Saif, M. (2008): Cycle


Adaptive Feedforward Approach Control of an
Electromagnetic Valve Actuator. Proceedings of
the 47th IEEE Conference on Decision and
Control, 5698-5703.

Keilig, R. (2007): Entwurf von


schnellschaltenden (hochdynamischen) neutralen
Elektromagnetsystemen. Dissertation TU Ilmenau.

Costagliola, M. (1950): The Theory for Spring


Loaded Valves for Reciprocating Compressors. J.
Appl. Mech, 415-420.
9

Aigner, R. Meyer, G., Steinrck H. (2005): Valve


Dynamics and Internal Waves in a Reciprocating
Compressor. 4th EFRC-Conference, 169-178.
10

Aigner, R. (2007): Internal Flow and Valve


Dynamics in a Reciprocating Compressor.
Dissertation TU Vienna.

11

Workshop documentation of the 5th EFRCConference (2007): Capacity Control.

10/10

Anda mungkin juga menyukai