U.S. patent application number 12/689491 was filed with the patent office on 2010-08-12 for high pressure pump control.
This patent application is currently assigned to Waters Technologies Corporation. Invention is credited to Jose de Corral, Stanley P. Pensak, JR..
Application Number | 20100202897 12/689491 |
Document ID | / |
Family ID | 35839723 |
Filed Date | 2010-08-12 |
United States Patent
Application |
20100202897 |
Kind Code |
A1 |
Corral; Jose de ; et
al. |
August 12, 2010 |
HIGH PRESSURE PUMP CONTROL
Abstract
A feedback control loop for a high pressure pump modifies the
accumulator velocity and pressure during solvent transfer. The
accumulator velocity is adjusted to maintain the system pressure
equal to the expected pressure to thereby eliminate the effect of
the flow deficit created by a thermal effect.
Inventors: |
Corral; Jose de; (Grafton,
MA) ; Pensak, JR.; Stanley P.; (East Walpole,
MA) |
Correspondence
Address: |
EDWARDS ANGELL PALMER & DODGE LLP
P.O. BOX 55874
BOSTON
MA
02205
US
|
Assignee: |
Waters Technologies
Corporation
Milford
MA
|
Family ID: |
35839723 |
Appl. No.: |
12/689491 |
Filed: |
January 19, 2010 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11631354 |
Oct 22, 2007 |
|
|
|
PCT/US05/24108 |
Jul 6, 2005 |
|
|
|
12689491 |
|
|
|
|
60587381 |
Jul 13, 2004 |
|
|
|
Current U.S.
Class: |
417/53 ;
417/521 |
Current CPC
Class: |
F04B 49/065 20130101;
F04B 2205/01 20130101; F04B 2205/11 20130101; F04B 11/0058
20130101; F04B 2205/09 20130101; G01N 2030/326 20130101; F04B 23/06
20130101; F04B 2205/05 20130101; F04B 11/0075 20130101; F04B
2205/10 20130101 |
Class at
Publication: |
417/53 ;
417/521 |
International
Class: |
F04B 49/08 20060101
F04B049/08; F04B 41/06 20060101 F04B041/06 |
Claims
1-40. (canceled)
41. A flow control system for controlling deliver of a high
pressure solvent in a chromatography system at pressures at which
the effects of adiabatic heating become noticeable, the flow
control system comprising: a high pressure pump for delivering the
solvent, the high pressure pump having two interconnected pump
heads with a first piston of a first pump and a second piston of a
second pump; a pump controller for setting a nominal velocity
profile of the second piston which during transfer is substantially
a mirror image of the first piston profile; and a closed loop
feedback control on a pressure of at least one of the first and
second pumps, the closed loop feedback control being operative
during a control period for compensating for a deficit in solvent
delivery caused by adiabatic heating during solvent compression by
modifying at least one of the nominal velocity profile of the
second pump or a nominal velocity profile of the first pump during
solvent transfer such that a system pressure is maintained
substantially equal to a desired target pressure.
42. The flow control system according to claim 41, wherein the
control period is substantially equal to a period of transfer of
the solvent.
43. The flow control system according to claim 41, wherein the
control period is longer than a period of transfer of the
solvent.
44. The flow control system according to claim 41, wherein the
control period is shorter than a period of transfer of the
solvent.
45. The flow control system according to claim 41, further
comprising a second piston transducer for measuring the system
pressure.
46. The flow control system according to claim 41, further
comprising a first pump pressure transducer for measuring a
pressure within a chamber of the first pump.
47. The flow control system according to claim 41, further
comprising a fluidic restrictor between an outlet of the high
pressure pump and a solvent load of the chromatography system.
48. The flow control system according to claim 47, wherein the
fluidic restrictor is a portion of narrow inner diameter
tubing.
49. The flow control system according to claim 41, further
comprising a filter for removing the second pump pressure of the
high pressure pump to eliminate frequencies beyond a bandwidth of
the closed loop feedback control.
50. The flow control system according to claim 41, wherein the
closed loop feedback control is based on a PID controller.
51. The flow control system according to claim 41, further
comprising a digital signal processor for performing the
modification of the second piston velocity.
52. The flow control system according to claim 41, wherein the
second piston nominal velocity profile is modified by reducing a
velocity of the second piston.
53. The flow control system according to claim 41, wherein the flow
control system is adapted for controlling a further high pressure
pump with interconnected pump heads and wherein both high pressure
pumps are connected in parallel and each high pressure pump has a
pump cycle.
54. The flow control system according to claim 41, further
comprising means for monitoring position of the pistons of the
first and second pumps in order to avoid control periods
overlapping.
55. The flow control system according to claim 41, further
comprising means for interchanging data relating to respective
positions within a pump cycle between both pumps in order to avoid
control periods overlapping.
56. The flow control system according to claim 41, further
comprising means for advancing a control period with a longer pump
cycle for one of the pumps when a control period collision is
foreseen to thereby substantially avoid overlap with a control
period of the other pump.
57. The flow control system according to claim 41, wherein the
first and second pump are connected in series.
58. The flow control system according to claim 41, wherein the
first and second pump are connected in parallel.
Description
RELATED APPLICATIONS
[0001] This application claims the benefit of U.S. Provisional
Application No.: 60/587,381 filed Jul. 13, 2004, the entire
contents of which are incorporated herein by reference.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The subject invention is directed generally to a scientific
laboratory analytical equipment, and more particularly, to analytic
equipment with a closed loop feedback controller for a high
pressure pump.
[0004] 2. Background of the Related Art
[0005] Scientific laboratories commonly need to separate chemical
compounds on such basis as the compounds molecular weight or size,
charge or solubility. Separation of the compounds is often a first
step in the identification, purification and quantification of the
compounds. Chromatography or, more specifically, high performance
liquid chromatography (HPLC) has become the analytical tool of
choice for applications as varied as biotechnological, biomedical,
and biochemical research as well as for the pharmaceutical,
cosmetics, energy, food, and environmental industries.
[0006] As advances in technology emerge, manufacturers of HPLC
instruments are quick to improve the performance of their product
lines. In fact, improvements in one technological area or subsystem
typically spurn on advancement in interrelated areas or subsystems.
U.S. Pat. No. 6,187,595 to Staal, which is incorporated herein by
reference in its entirety, discusses several advantages and
disadvantages related to evolving approaches based on new
technology.
[0007] Currently, there are several types of pumps commonly used as
subsystems with HPLC instruments. HPLC instruments may incorporate
reciprocating pumps, syringe pumps, and constant pressure pumps as
are known to those of ordinary skill in the art. For example, most
reciprocating pumps include a small motor driven plunger which
moves rapidly back and forth in a hydraulic chamber to vary the
volume thereof. On the back stroke, the plunger pulls in a solvent.
On the forward stroke, the plunger of the reciprocating pump pushes
the solvent out to a column. In order to achieve the desired flow
stability within the column, multiple plungers are employed,
normally two. The two plungers may be employed in series or in
parallel to achieve the desired delivery flow and pressure.
[0008] During compression of the solvent, energy is absorbed that
raises the temperature of the solvent. This thermal effect is
proportional to the solvent compressibility, the target pressure
(e.g., the desired instrument pressure) and the rate at which the
solvent is compressed. For many leading edge technology HPLC
instruments, the high pressure and limited amount of time to
compress the solvent creates significant thermal effect. The heat
is usually dissipated to the surroundings and associated instrument
at a rate dependent upon the relative mass and thermal conductivity
of the compressed solvent and the surroundings. In most
applications, for pressures up to a couple thousand psi, the
thermal effects of compression are negligible.
[0009] However, the thermal effects at high pressure become more
appreciable. The thermal effects create errors in the pressure of
the compressed solvent because the solvent temperature is elevated
during compression compared with during analysis in the instrument.
In other words, just after the solvent is compressed to the target
pressure, the pressure decays as the solvent temperature moves
toward equilibrium with the instrument. As a result, the compressed
solvent settles to a pressure below the target pressure and,
thereby, creates a deficit in delivered flow.
[0010] Prior art pump control systems lack the required ability to
react to the thermal effects of solvent compression at high
pressures. So despite the advances of the state of the art, HPLC
instruments are lacking in stability and performance. As a result,
inaccurate results are still common. Moreover, such prior art
instruments are plagued by inadequacies such as complex electronics
and numerous additional components that undesirably increase costs
and complexity without overcoming the noted drawbacks. In view of
the above, it would be desirable to provide a controller for a high
pressure pump that affords accurate delivery of the target pressure
and the ability to compensate for thermal effects.
SUMMARY OF THE INVENTION
[0011] The subject invention provides a controller for a high
pressure pump of the series type, wherein the feedback is applied
only when the primary plunger delivers flow, just after solvent
compression. Tile feedback ceases when the primary plunger ends
flow delivery therefrom, and the pump continues with the normal
flow delivery.
[0012] It is an object of the present invention to prevent solvent
composition errors when the flow from two pumps is used to create a
solvent gradient by correcting the delivered flow deficit occurring
after solvent compression.
[0013] It is understood that this flow deficit is created by
adiabatic heating that occurs when the solvent is compressed, and
it is proportional to the solvent compressibility, compression
pressure, rate at which the solvent is compressed, and thermal mass
of the compressed solvent relative to the pump head thermal
mass.
[0014] It is an object of the present invention to provide a
controller for a high pressure pump that uses continuous closed
loop feedback on the delivered solvent so that adjustment occurs to
maintain the instrument pressure at the target value.
[0015] It is an object of the present invention to provide a
controller for a high pressure pump that eliminates flow deficits
caused by the thermal effects created during compression of the
solvent.
[0016] It is another object to provide a controller for a high
pressure pump that can compensate for thermal effects created
during compression of the solvent.
[0017] It is still another object to provide a controller for a
high pressure pump that achieves quick and accurate response to
dynamic flow conditions.
[0018] The foregoing objects are achieved by the instant invention
which, in one aspect, provides a flow control system for
controlling a high pressure pump for delivering a fluid load having
a primary piston and an accumulator piston, each piston having a
velocity and pressure associated therewith. The flow control system
comprises a closed loop feedback control on the accumulator
pressure during transfer for modifying the accumulator velocity to
maintain a system pressure substantially equal to the expected
pressure.
[0019] In another aspect, the invention provides a method for
controlling an output of a high pressure pump system to reduce an
effect created by adiabatic heating, wherein the high pressure pump
system includes a primary piston and an accumulator piston, each
piston having a velocity and pressure associated therewith. The
method comprises the step of modifying the accumulator velocity
with a closed loop feedback control on the accumulator pressure
during transfer to maintain a system pressure substantially equal
to an expected system pressure.
[0020] In yet another aspect, the invention provides a method for
controlling a velocity of a primary piston in a high pressure pump,
wherein an accumulator piston delivers solvent to a system and the
primary piston refills the accumulator piston and delivers solvent
to the system while the accumulator piston is refilling. The method
comprises the step of controlling the velocity of the primary
piston before transfer by instructing the primary piston to
compress the solvent some time before transfer is due such that
adiabatic heating effects extinguish before transfer starts.
[0021] In still another aspect, the invention comprises a
computer-readable medium whose contents cause a control system to
perform a method for controlling an output of a high pressure pump
system to eliminate delivery error, wherein the high pressure pump
system includes a primary piston and an accumulator piston, each
piston having a velocity and pressure associated therewith. The
control system has a digital signal processor and a program with
functions for invocation by performing the steps of modifying the
accumulator velocity with a closed loop feedback control on the
accumulator pressure during transfer to maintain a system pressure
substantially equal to an expected system pressure.
[0022] Another aspect of the invention provides a method for
controlling an output of a high pressure solvent pump system to
reduce delivery error, wherein the high pressure pump system
includes a primary piston and an accumulator piston, each piston
having a velocity and pressure associated therewith. The method
comprises the steps of: periodically applying a closed loop
feedback control having a closed loop bandwidth to maintain a
system pressure substantially equal to an expected system pressure;
and filtering the system pressure to eliminate frequencies beyond
the closed loop bandwidth.
[0023] In another aspect ,the invention provides a system for
controlling an output of a high pressure pump system to reduce
error in solvent delivery. The system comprises a high pressure
pump including a primary piston and an accumulator piston, each
piston having a velocity and pressure associated therewith; a
closed loop feedback control for modifying the accumulator velocity
on the accumulator pressure during transfer to maintain a system
pressure substantially equal to an expected system pressure; and
means for restricting fluid between an outlet of the high pressure
pump and a fluid load of the system.
[0024] In yet another aspect, the invention provides a method for
computing a pressure set point to follow an expected pressure trace
in a flow control system for controlling a high pressure pump
having a primary piston and an accumulator piston, each piston
having a velocity and pressure associated therewith, the flow
control system having a closed loop feedback control on the
accumulator pressure during a control period for modifying the
accumulator velocity to maintain a system pressure substantially
equal to the expected pressure. The method comprises the steps of
:
[0025] computing an initial set point based upon using pressure
values immediately before the closed loop feedback control
activates;
[0026] computing changes to the set point during the control
period; performing a prediction of the set point based upon the
initial set point and the previous pressure values; and
[0027] projecting pressure values during the control period to
determine the pressure set point.
[0028] In still another aspect, the invention provides a flow
control system for controlling a high pressure pump having a
primary piston and an accumulator piston, each piston having a
velocity and pressure associated therewith. The flow control system
comprises a closed loop feedback control on the accumulator
pressure during a control period for modifying the accumulator
velocity to maintain a system pressure substantially equal to the
expected pressure.
[0029] Another aspect of the invention provides a method for
protecting a pump by avoiding a possibility that a pressure control
mechanism could command a piston within the pump beyond actual
capabilities of the pump. The method comprises the steps of:
[0030] computing a volume over a nominal delivery that has been
delivered by the pressure control mechanism; and
[0031] deactivating the pressure control mechanism if the volume
delivered exceeds a threshold value.
[0032] Yet another aspect of the invention provides a method for
isolating a control loop from external fluid conditions such that
oscillation from connecting a pair of pumps in parallel is
substantially prevented, each pump having a pump cycle, the method
comprising the steps of
[0033] interchanging data related to respective positions within a
pump cycle between the pair of pumps in order to substantially
avoid overlap of respective control periods; and
[0034] advancing a control period of the pump with longer pump
cycle when a control period collision is foreseen to thereby
substantially avoid overlap with a control period of the other
pump. In one embodiment of this aspect, the method further
comprises the steps of
[0035] periodically applying a closed loop feedback control having
a closed loop bandwidth to maintain a system pressure substantially
equal to an expected system pressure;
[0036] filtering the accumulator pressure to eliminate frequencies
beyond the closed loop bandwidth;
[0037] controlling an output of a high pressure solvent pump system
to reduce delivery error, wherein the high pressure pump system
includes a primary piston and an accumulator piston, each piston
having a velocity and pressure associated therewith; and
[0038] modifying the accumulator velocity with the closed loop
feedback control on the accumulator pressure during transfer.
[0039] It should be appreciated that the present invention can be
implemented and utilized in numerous ways, including without
limitation as a process, an apparatus, a system, a device, a method
for applications now known and later developed or a computer
readable medium. These and other unique features of the system
disclosed herein will become more readily apparent from the
following description and the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
[0040] So that those having ordinary skill in the art to which the
subject invention appertains will more readily understand how to
make and use the same, reference may be had to the figures
wherein:
[0041] FIG. 1 is a schematic view of a high pressure serial pump
constructed in accordance with the subject invention.
[0042] FIG. 2 illustrates typical profiles of velocity and pressure
in both pistons, during an entire pump cycle of a preferred
embodiment.
[0043] FIG. 3 illustrates an example of the flow deficit and the
correspondent pressure dip, relative to the velocity profiles
during transfer of a preferred embodiment.
[0044] FIG. 4 illustrates the difference between the expected and
actual accumulator (system) pressure traces of a preferred
embodiment.
[0045] FIG. 5 shows how the accumulator velocity is modified during
the flow deficit, from its nominal velocity profile of a preferred
embodiment.
[0046] FIG. 6 shows an example of how a linear prediction is used
to compute the pressure set point based on two pressure values just
before the start of control of a preferred embodiment.
[0047] FIG. 7 illustrates a simplified block diagram of the control
loop of a preferred embodiment.
[0048] FIG. 8 illustrates typical profiles of velocity and pressure
in both pistons, during an entire pump cycle of another preferred
embodiment.
[0049] FIG. 9 illustrates the effect of the pressure control on the
primary pressure, and how the primary velocity is modified by the
control loop.
DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS
[0050] The present invention overcomes many of the prior art
problems associated with controlling high pressure pumps. The
advantages, and other features of the system disclosed herein, will
become more readily apparent to those having ordinary skill in the
art from the following detailed description of certain preferred
embodiments taken in conjunction with the drawings which set forth
representative embodiments of the present invention.
Solvent Delivery System Background
[0051] Referring to FIG. 1, a schematic view of a high pressure
serial pump is illustrated. High pressure pumps for use in
chromatography applications normally use a reciprocating type of
design involving two pistons. Depending on the fluidic
configuration, there are two main design types: parallel or series.
In the parallel design, the two pistons alternate in operation
whereby one piston delivers flow while the other intakes new
solvent from the solvent source. In the series design, only one
piston intakes solvent from the solvent source (the primary), while
the other (the accumulator) does most of the solvent delivery. The
primary takes the responsibility of refilling the accumulator at
high pressure when, inevitably, the accumulator needs to intake new
solvent.
[0052] Preferably, a pressure control algorithm is used in a series
pump design; thus for simplicity only the series pump will be
covered here.
[0053] The primary piston intakes solvent from the solvent source
and delivers the solvent to the accumulator piston. The
accumulator, then, delivers the solvent to the system. The check
valves are passive valves that allow fluid to go in one direction
only. Respective pressure transducers measure the pressure at the
outlet of each piston. All components are in the same flow path
forming a series fluid circuit; thus the name "series". Normally,
the plunger size is the same for both pistons. While the
accumulator is delivering flow to the system at high pressure, the
primary intakes new solvent from the solvent source and waits until
it is time to refill the accumulator. At that point, the primary
compresses the solvent to the same pressure measured by the
accumulator transducer (the system pressure), and is set ready for
delivering to the accumulator. When the accumulator reaches the end
of its delivering motion, the pump controller (not shown) instructs
the primary to start delivering and the accumulator to start
intaking. This operation, known as "transfer", is done at high
pressure and continues until the accumulator is completely full and
ready to resume its normal delivery.
[0054] While transfer is taking place, the accumulator is,
obviously, not delivering to the system. Therefore, the primary has
to take over that responsibility in order to avoid interruption in
the flow delivered to the system. To accomplish this task, transfer
is done at a much higher plunger velocity than the accumulator's
normal delivery velocity, and a portion of the primary delivery
goes to the system.
[0055] Once transfer is finished, the pump controller instructs the
accumulator to resume normal delivery, and the primary to intake
new solvent. This cycle, known as the "pump cycle", repeats
continuously while the pump is delivering solvent to the system.
The pump cycle duration depends mainly on the stroke volume and the
delivered flow. The roll of the check valves is easy to understand.
The primary check valve allows the primary to intake solvent at
atmospheric pressure from the solvent source, but prevents the
solvent from going back to the solvent container when the primary
compresses the solvent to the system pressure. The accumulator
check valve allows the primary to deliver solvent to the
accumulator, but prevents the accumulator delivery at high pressure
from going back to the primary when the primary is intaking new
solvent at atmospheric pressure.
[0056] The accumulator pressure transducer measures the system
pressure, and provides the input to the pressure control algorithm.
The accumulator also provides the target compression pressure to
the primary, when the primary starts the compression of new
solvent. The primary pressure transducer measures the pressure
inside the primary, so the compression is stopped when the pressure
reaches the compression target. Referring to FIG. 2, typical
profiles of velocity and pressure in both pistons, during an entire
pump cycle is shown.
High Pressure Compression Effects
[0057] When the solvent inside the primary is compressed, its
temperature rises. This temperature increase, known as adiabatic
heating, is lost to the solvent surroundings and to the system
(when the primary starts delivering), at a rate dependent on the
relative mass and thermal conductivity of the compressed solvent
and the surroundings.
[0058] However, this temperature increase creates an error in the
pressure of the compressed solvent, because the solvent temperature
at the time of compression is higher than the temperature the
solvent will eventually have (the temperature of the system).
[0059] Therefore, just after the solvent is compressed to the
target pressure (the system pressure), its pressure starts to decay
as its increased temperature starts to be equilibrated . to the
system temperature. The compressed solvent pressure eventually
settles at a value below the intended system pressure. This creates
a deficit in delivered flow when the primary starts delivering.
[0060] The thermal effect is proportional to the solvent
compressibility, to the compression pressure (the system pressure),
and to the rate at which the solvent is compressed. For pressures
up to a few thousand psi this thermal effect can normally be
ignored, but it is significant at higher pressures. Furthermore,
due to the timing involved in the reciprocating pumps' action,
there is normally a limited amount of time to compress the solvent
from atmospheric pressure to system pressure. Therefore, this
thermal effect creates significant flow delivering errors, which
represent solvent composition errors when the solvents of two pumps
are combined together at high pressure to form a solvent
gradient.
Pressure Control Description
[0061] Referring to FIG. 3, an example of the flow deficit and the
correspondent pressure dip, relative to the velocity profiles
during transfer is shown. As the primary is where the adiabatic
heating effect takes place, the flow deficit occurs when the
primary enters in fluid communication with the accumulator. Up to
that point, the pump flow is correct, as the pump flow is delivered
by the accumulator only.
[0062] The flow deficit shows up as a pressure dip and is sensed by
the accumulator pressure transducer. As the fluidic dynamics
affects the measured pressure, the pressure dip profile does not
necessarily match the flow deficit profile. They would only match
if the time constant of the fluidics is smaller than the time
constant of the flow deficit (small fluidic capacitance or small
fluidic resistance).
[0063] The flow deficit duration varies mainly with pressure and
solvent compressibility and normally lasts for a period shorter
than transfer, although the flow deficit duration could extend
beyond the duration of transfer.
[0064] Referring to FIG. 4, the difference between the expected and
actual accumulator (system) pressure traces is shown. To compensate
for this error, the accumulator velocity is modified during the
flow deficit, from its nominal velocity profile, as shown in FIG.
5. The modification of the accumulator velocity is done with a
closed loop feedback control on the accumulator pressure during
transfer. The accumulator velocity is adjusted to maintain the
system pressure equal to the expected pressure. This eliminates the
effect of the flow deficit created by the thermal effect.
[0065] As mentioned above, the pressure control runs when the
primary is delivering (during transfer), and ceases when the
primary ends its delivery. Once transfer is finished, the pump
continues with the normal accumulator delivery. Therefore, the
feedback control is applied only once per pump cycle. The feedback
control does not run continuously, i.e., the feedback control can
be turned on before transfer, and turn off after transfer.
[0066] The intermittent use of the feedback control yields improved
delivery accuracy. Preferably, the feedback control is only
utilized enough to achieve the desired accuracy. During transfer,
as the figures above show, the accumulator nominal velocity profile
is set by the pump controller, and it is substantially a mirror
image of the primary velocity profile. The pressure control
algorithm adjusts the accumulator velocity on top of this nominal
velocity profile, unaware that the pump controller is modifying the
accumulator velocity too. The pressure control does not modify the
primary velocity.
Pressure Control Algorithm
Pressure Filter
[0067] Before using the measured pressures for pressure control,
they should be filtered to eliminate frequencies beyond the closed
loop bandwidth. If these frequencies are too close to a multiple of
the loop sampling frequency, the frequencies could alias into the
control bandwidth, creating unexpected loop behavior.
[0068] The main source of high frequency components in the measured
pressures is the resonance of the pump motors, which is about 200
Hz. There is also high frequency noise associated with the pressure
transducers. The electronic two pole Butterworth filter at 225 Hz,
located just before the analog to digital (A/D) conversion, does
not filter these frequencies enough.
[0069] A software digital filter running in the pump controller
digital signal processor (DSP) is used to remove the high
frequencies in the measured pressures.
[0070] The filter runs at the A/D converter rate (2.441 KHz), and
is made of two cascaded single pole IIR filters at 100 Hz, followed
by a sync filter (FIR filter) that averages the last 12 samples.
The sync filter effectively removes most of the motor resonance
frequency. The filter creates a lag of about 5 milliseconds.
Control Start And End Points
[0071] In a preferred embodiment, the pressure control occurs
advantageously during transfer. In alternative embodiments, the
period of pressure control may be longer or shorter than transfer.
For example, the pressure control may start Wore transfer begins,
and extends until sometime after transfer is finished. The start
point is set at the point when the primary starts to compress the
solvent, and the end point is about 50 milliseconds after the
solvent inside the primary has been decompressed. This extended
control period allows compensation for other flow delivery errors
that occur around transfer, due to mechanical imperfections and the
like.
Fluidic Isolation
[0072] In order to make the feedback control loop work optimally
under different fluid load conditions, it is desirable to add a
small amount of fluid restriction between the outlet of the pump
and the system fluid load. In a preferred implementation, this
fluid restriction is created with 12'' of 0.005'' inner diameter
(ID) tubing. This decouples the dynamic load seen by the feedback
loop from the external fluid capacitance, which, otherwise, would
need to be computed for each load condition.
[0073] This fluid restriction also helps isolate the effect that
one pump creates on the other, when two pumps are connected in
parallel to fowl high pressure solvent gradients. Without isolation
restrictions, the feedback control loop would oscillate when the
control periods of both pumps overlap. This is because both pumps
would try to compensate the same pressure, which is created by
contributions from both pumps. Each control loop would not know
which is the portion it has to compensate.
[0074] The isolation restrictors allow each control loop to get a
pressure measurement that is dominated by its own contribution,
although it contains some amount created by the other pump. The
restrictors along with the system fluid capacitance create enough
isolation at the loop crossover frequency, sufficient to prevent
oscillation.
Pressure Set point Computation
[0075] Another element of this control algorithms the computation
of the correct control loop set point at any given time during the
control period. The pressure set point is not necessarily constant
during the entire control period. Actually, the pressure set point
should follow the expected pressure trace that would have been if
the pressure dip was not present.
[0076] The pressure set point algorithm computes the initial set
point and how the actual set point changes during the control
period, using the pressure values immediately before the control
starts. The pressure set point algorithm does a linear prediction
with these pressure values, and projects the pressure values during
the control period. These projected values are used as the pressure
set point. A linear prediction has proved to be 3.5 sufficient when
the control period is less than one second. However, curved
prediction methods could be used as well, such as a quadratic
fit.
[0077] FIG. 6 shows an example of how a linear prediction is used
to compute the pressure set point based on two pressure values just
before the start of control. The two pressure values are used to
generate a straight line that determines the set point values. The
figure shows two cases, one with a horizontal prediction, and the
other with a sloped prediction.
[0078] The set point algorithm uses a strongly filtered version of
the pressure to perform the linear prediction. This digital filter
provides the trend of the pressure, and prevents local noisy
pressure samples from affecting the correct set point or set point
slope computation. This digital filter is a single pole low pass
filter with a corner frequency that varies with the set-flow,
between 10 Hz for the highest flow, and 0.01 Hz for the lowest
flow.
[0079] The algorithm also takes into account the set-flow changes,
in order to compute the correct set point slope under varying fluid
load conditions.
Feedback Control Loop
[0080] The pressure feedback control is designed as a
proportional-integral-derivative (PID) control loop. The inputs are
the pressure set point and the filtered accumulator pressure, and
the output is the accumulator velocity. The loop sampling frequency
is set to 200 Hz.
[0081] A simplified block diagram of the control loop is shown in
FIG. 7 with Table 1 serving as a legend.
TABLE-US-00001 TABLE 1 Press Actual pressure (psi) Kp Conversion
constant: actual pressure (counts)/(psi) to measured pressure P
Measured pressure (counts) HF(s) Filter transfer function Pf
Filtered pressure (counts) Pset Pressure set point (expected
pressure (counts) without the pressure dip) Pe Pressure error
(counts) Hc(s) Compensation transfer function V Accumulator
velocity (steps/sec) Kv Conversion constant: velocity to flow
(ml/min)/(steps/sec) Q Actual delivered flow (ml/min) HFL(s)
Fluidics transfer function
[0082] The formulas based upon this exemplary embodiment are as
follows:
Kp = 2.64254 ( counts psi ) ##EQU00001## H F ( .omega. ) = 1 ( 1.59
10 - 3 j .omega. + 1 ) 2 1 - exp ( - j.omega. 12 .omega. s ) 1 -
exp ( - j.omega. 1 .omega. s ) 12 ( counts counts ) ##EQU00001.2##
.omega. s = 15337 ( rad / sec ) = filter sampling frequency
##EQU00001.3## H C ( s ) = Kc ( .tau. 1 s + 1 ) ( .tau. 2 s + 1 ) s
( steps / sec counts ) ##EQU00001.4## Kc = compensation gain
##EQU00001.5## .tau. 1 = first compensation zero time constant
##EQU00001.6## .tau. 2 = second compensation zero time constant
##EQU00001.7## Kv = 2.3538 10 - 4 ( ml / min steps / sec )
##EQU00001.8## H FL ( s ) = ( .tau. F s + 1 ) R i + R F ( .tau. F
.tau. i ) s 2 + ( .tau. F + .tau. i + .tau. c ) s + 1 ( psi ml /
min ) ##EQU00001.9## .tau. F = R F C F ##EQU00001.10## .tau. i = R
i C H ##EQU00001.11## .tau. c = R F C H ##EQU00001.12## R F = fluid
load resistance ##EQU00001.13## C F = fluid load capacitance
##EQU00001.14## R i = fluid isolation resistance ##EQU00001.15## C
H = fluid pump head capacitance ##EQU00001.16##
[0083] For the typical fluid conditions, the loop crossover
frequency is 14 Hz with a phase margin of 61 degrees, and a
settling time of 56 milliseconds. This loop behavior does not
change much for a wide range of fluid conditions.
[0084] The fluid pump head capacitance is a gain term in the loop,
and its value is determined for each type of solvent being used.
This capacitance is computed in the primary when the solvent is
compressed to the target pressure.
Feed Forward Compensation
[0085] This step of the control algorithm intends to compensate for
pressure errors with frequencies beyond the loop crossover
frequency, such as those created by certain types of check valves.
These pressure disturbances are too fast and the control loop
cannot compensate for them.
[0086] First, the high frequency components of the pressure are
separated using a digital high pass filter with a corner frequency
above the loop crossover frequency. Then, an additional
compensation velocity is added to the accumulator velocity,
computed as follows.
[0087] The feed forward accumulator velocity contribution is a
factor of the set-velocity. This factor is inversely proportional
to the ratio of the high frequency pressure components and the
pressure set point.
Control Guard
[0088] This is a protection algorithm to avoid the possibility that
the pressure control commands an accumulator velocity or position
beyond the actual capabilities of the pump mechanics.
[0089] In this regard, an important control situation is the
section between the start of control, and the start of transfer. At
this point, the accumulator is near the end of its possible
displacement, and any exaggerated velocity increase requested by
the control loop will most likely result in the accumulator plunger
hitting the hardware stop. For example, an air bubble entering the
pump is a typical scenario that could lead to this situation. The
pressure control will increase the accumulator velocity
substantially trying to compress the bubble.
[0090] This is not the case once transfer has started, because the
pump controller has initiated the accumulator intake. The intake
velocity is higher than the highest possible velocity set by the
control loop, so it is not possible for the plunger to hit the
hardware stop.
[0091] The control guard algorithm computes the volume over the
nominal delivery that has been delivered by the control loop so
far, and turns the control off if the volume delivered exceeds a
reasonable threshold value.
Two Pump Control Collision Avoidance
[0092] The isolation restrictors, discussed above, isolate the
control loop from external fluid conditions, and prevent the
control loop from oscillating when two pumps connected in parallel
overlap their control periods. However, this isolation is not
enough for high precision solvent gradients, where the small
remaining interaction between both pump's control loops creates
solvent composition errors.
[0093] To eliminate these errors, the two pumps interchange
information about their respective position within the pump cycle,
in order to avoid the control periods overlapping. In other words,
each pump knows the other pump's cycle duration and current
position within that cycle.
[0094] When a control period collision is foreseen, the pump with
longer pump cycle advances its control period just enough to avoid
the overlap with the other pump control period. This technique
effectively removes any remaining composition errors in solvent
gradients.
Alternative Pressure Control Embodiment
[0095] In another embodiment, the pressure control algorithm is
based on controlling the primary velocity before transfer, rather
than the accumulator velocity during transfer. The nominal primary
velocity profile is changed slightly.
[0096] In the standard serial pump design, as described above, the
primary does not compress the solvent until transfer is just about
to occur. When the primary compresses the solvent, the adiabatic
heating effect takes place, and transfer starts. This is the reason
why the flow deficit occurs during transfer.
[0097] However, the primary can be instructed to compress the
solvent some time before transfer is due, to let the adiabatic
heating effect extinguish by the time transfer starts. FIG. 8 shows
the pressure and velocity profiles with this change (compare with
similar FIG. 2).
[0098] FIG. 8 shows how the adiabatic heating effect creates a
pressure decay on the primary pressure just after compression. This
pressure decay will create a flow deficit on the delivered flow
when transfer starts.
[0099] This alternative control approach deals with the adiabatic
heating effect problem by controlling the primary pressure between
the end of compression and the start of transfer. The pressure
control adjusts the primary velocity to maintain the primary
pressure slightly below the accumulator pressure.
[0100] This compensates for the adiabatic heating effect and leaves
the primary pressure at the right value when transfer starts. The
primary pressure should be maintained below the accumulator
pressure to guarantee that the accumulator check valve stays closed
during the primary pressure control. Referring now to FIG. 9, the
effect of the pressure control on the primary pressure, and how the
primary velocity is modified by the control loop are shown.
[0101] There are inherent advantages of this alternative control
approach. An advantage is that the system is not affected by
external fluidic conditions, such as the fluidic load, or the
transfer overlap of a second pump connected in parallel. As the
accumulator check valve stays closed during the control period, the
primary pressure is not affected by anything happening downstream
of the accumulator check valve.
[0102] Also, there is no need for a set point computation
algorithm, because the accumulator pressure determines the set
point value.
[0103] In regard to disadvantages, this control alternative
provides compensation for the adiabatic heating effect only, and
not for the mechanical imperfections that affect transfer, for
which the other control algorithm compensates for.
Incorporation by Reference
[0104] All patents, published patent applications and other
references disclosed herein are hereby expressly incorporated
herein in their entireties by reference.
Equivalents
[0105] Although the subject invention has been described with
respect to preferred embodiments, those skilled in the art will
readily appreciate that various changes and/or modifications can be
made to the invention without departing from the spirit or scope of
the invention.
* * * * *