Due to the highly generated heat from cutting, milling and drilling
of high speed machining, the machine tool cooler is widely adopted for
heat rejection to keep the precise control of oil temperature. The machine
tool coolers are also the best managers of oil (or water) temperature
in avoiding the deviation of spindle centerline for machine tools. However,
most of the machine tool coolers are simply equipped with a simple ON/OFF
temperature controller. The temperature can merely be controlled within
the range of ± 3 ° C. For high-speed machine tool, such temperature
fluctuations will degrade the manufacturing performance significantly
and cause many side effects such as tool deformation, poor manufacturing
precision and reduction of lifetime for the machine tool. In addition,
an ON/OFF controlled compressor basically consumes much more energy due
to the frequent starting losses (Huang, 2001; Huang and Tsai, 2002). In
view of these shortcomings, an improved type of oil-cooling machine equipped
with a variable-frequency compressor together with a high performance
controller has been proposed to control the oil coolant with steady state
accuracy up to ± 0.3 ° C.
PID controllers are still applied extensively in industrial application.
A simple approach to obtain parameters of a PID controller for closed
loop system has been simulated and analyzed comprehensively (Kaya et
al., 2003). Besides, PID design method and future direction were presented
thoroughly by computerized and simulation-based approach (Li et al.,
2006). The differences between academic research and industrial practice
in PID controller have been discussed to motivate new research directions.
Moreover, a new design and tuning procedure for PID feedforward controller
is proposed to achieve a predefined process output transition time (Visioli,
2004). The proposed method represented better performance for inverse
model-based approach from simulation and experimental results. Furthermore,
a tuning algorithm for the PID controller utilizing fuzzy theory was conducted
by simulation (Hwang et al., 1999). The fuzzy auto tuning algorithm
for PID controller reduced the overshoot and rise time significantly.
Besides, the experiment of an intelligent refrigeration system utilizing
an inverter has been proposed to demonstrate the reducing of energy consumption
in comparison with the traditional on-off system under the same operating
condition (Buzelin et al., 2005). To control the refrigeration
system capacity, the fuzzy control of the compressor speed was applied
to evaluate the energy saving potential. A significant energy saving on
an average about 13% has been obtained using the compressor speed control
algorithm based on the fuzzy logic in comparison with the thermostatic
control (Aprea et al., 2004).
For the improved oil cooler, the motor speed can be continuously varied
to eliminate the frequent starting losses and the temperature tracking
control accuracy can be improved as well. However, since the oil-cooling
process is a rather nonlinear and time-varying dynamic behavior. Although
many existing control schemes such as PID control (Wang et al.,
1999) predictive control (Maciejowski, 2002) adaptive predictive control
have been proposed in literature, it is still a research topic to pursuit
a more practical, high performing and energy saving temperature controller.
In this study, a new PID control method is proposed for controlling the
temperature of an oil cooler system. The traditional PID control based
on the smith predictive method (Astrom et al., 1994) is adopted
as a basic controller of the cooler system. To deal with the nonlinear
time varying and time delay characteristic of the cooler system, the PWM
(pulse width modulation) technique is then chosen as an auxiliary control
to modulate the hot-gas by-pass solenoid valve with duty ratio to provide
the oil precise control of temperature. Not only simulation but also experimental
results are also given for demonstrating the effectiveness of the proposed
control scheme. The control scheme not only improves the performance of
PID controllers cost-effectively but also maintains a permanent temperature
control specific for machine tool coolers.
CONFIGURATION OF THE CONTROL SYSTEM
The basic configuration of the investigated oil cooler system is
shown in Fig. 1. There are two subsystems demonstrated
in Fig. 1. The first subsystem mainly involves the circulating
loop of the cooling oil circulating system. Through a fixed frequency
induction motor controlled pump, the oil is forced to circulate around
the loop containing the oil tank, pump, heat exchanger and the tool machine.
Another subsystem involves the circulating of the refrigerant system which
a variable frequency induction motor controlled compressor is installed.
The refrigerant of the cooler system circulates around the loop containing
compressor, condenser, refrigerant control valve and the heat exchanger
to remove away the heat from the machine tool.
As shown in Fig. 1, this system is controlled simply
through a variable Voltage Variable Frequency (VVVF) inverter to control
the induction motor speed, which could provide the precise control of
oil temperature at the outlet of the heat exchanger. This induction motor
is three-phase, 60 Hz, 114 V, 1.5 HP 4-pole squirrel cage type. Besides,
the inverter provides a three phase variable frequency and variable voltage
with a fixed voltage to frequency ratio to keep the air gap flux of the
induction motor constant. The input command of the inverter can be varied
from 0 V to 10 V corresponding to 0 rpm to 1800 rpm for the induction
||Basic configuration of the oil cooler system for machine
Furthermore, the proposed controller, based on the input signal of the oil temperature
at the outlet of the heat-exchanger, generates the desired output signal
as the inverter command signal.
PROPOSED NEW PID CONTROL SCHEME
Figure 2 represents the block diagram of the proposed
controller. The proposed controller basically consists of three parts.
They are PID controller, differential auxiliary circuit and PWM circuit,
respectively. The familiar PID controller based on the Smith method is
adopted as a basic controller of the system as shown in Fig.
3. However, utilization of this single controller is not sufficient
to achieve the desired performance specification. In fact, due to complicated
nonlinear and time varying characteristic of the cooler system, it is
quite challenging to provide precise temperature control if the controller
is not well tuned. To stabilize the complicated system, a Smith predictive
control scheme is added. The PWM technique is then chosen as an
auxiliary duty ratio control to modulate the by-pass solenoid value to
keep the oil temperature as constant as possible.
Since the time constant of the controlled system is rather long, traditional
PID controller is not good at controlling the systems with large delay
time. It is quite easy to cause the temperature fluctuations for the cooler
system. In the larger delay systems, the pure delay segment e-τs
is the source term which leads to the confusion of the control information.
The control information can approach the controller only after a long
period of time because of the delay term (e-τs). Even
worse, the controller will not work appropriately when the delay term
becomes too large. It is quite essential and critical to adjust the control
information to improve the controller`s performance of controller.
Thereafter, the Smith compensate arithmetic algorithm with Smith controller
which was specially designed for large delay system was used to solve
the problem mentioned previously.
||Block diagram for the proposed controller
||Block diagram of the Smith predictor
The first order delayed model transfer function from literature (Mudi and Pal, 1999) was adopted
and can be expressed as:
The transfer function of system is G(s) e-τs, so the
Smith compensation is G(s) (1-e-τs). The equivalent diagram
of the Smith predictor is showed in Fig. 4. The control
section of Smith predictor was shown in Fig. 4. The
transfer function of the enclosed dash line area is GPID. The
PID controller was conducted by exploiting the system identification model
of Eq. 1 and match with the simple formula derived as:
where, T, T`, K and τ represent time constant, expectative time constant,
gain and delay time of system model, respectively, whereas Kp is
proportional constant, Ti is integral time constant, Td
is the derivative time constant and
is the transfer function of PID controller
based on e-τs equal to (1-τs) approximately.
||Equivalent diagram of the Smith predictor
||Measured oil temperature for system identification using
a step input to the controlled system under open loop condition
Under the appropriate design of the controller, we can fulfill the accuracy of control system merely
by a simple comercial available PID controller.
In this study, the step input (10 V) of signal is applied to the inverter.
The oil temperature at the outlet of the exchanger is measured and represented
in Fig. 5. The on-site measurement data can be used
to conduct system identification and to estimate the coefficients of G(s).
In other words, the model G(s) is chosen according to system identification
based on measurement data and can be expressed as the following:
RESULTS AND DISCUSSION
To demonstrate the effectiveness of temperature control for the
proposed control scheme, both simulation and experimental results are
presented. The traditional PID controller based on the Ziegler-Nichols
method is first determined for comparison. For the measured system after
system identification, the corresponding parameters of the traditional
PID controller have been determined as Kp = 4, Ti
= 100, Td = 28.5. The oil temperature at the outlet of the
heat exchanger was kept at 24 ° C in the initial stage (t = 0 sec).
The set points were chosen to be 22 ° C under constant loading of 1000
||Comparison of the traditional PID controller and the
Comparison for the simulation results of the traditional PID control
and the proposed control are shown in Fig. 6. The temperature
fluctuation of traditional PID controller is much lager than that of proposed
method. The control performance parameters such as maximum overshoot,
rise time, setting time and steady state error are far from the proposed
control scheme. It also reveals that application of traditional PID controller
still exist some problems in dealing with the nonlinear time varying and
time delay characteristic of the cooler system for high speed machining
(Fig. 6). The performance of traditional PID controller
is obviously not good enough for our goal. As shown in Fig.
6, the control accuracy of the proposed control scheme is better than
that of traditional PID controller and can achieve the precise temperature
control within ± 0.3 ° C for the steady state. Furthermore,
the proposed control scheme can provide satisfactory result even under
the effect of abrupt load disturbance. Therefore the controller developed
by our study is pretty simple and cost-effective, also can be easily adopted
by industry application.
As shown in Fig. 7, one can see that the temperature
fluctuation is much lager than that of Fig. 8. The performance
of the traditional PID control is obviously not good enough. However,
for the proposed control schemes (Fig. 8), the steady
state error is within ± 0.3 ° C. The experimental results reveal
that the satisfactory control accuracy of temperature in spite of the
nonlinear time varying characteristic of the plant.
From Eq. 2, the proportional constant Kp
can be obtained from the coefficient of time constant (T), expectative
time constant (T`), gain (K) and delay time (τ) of system model. Actually the integral time constant (Ti) is equal to the coefficient
of time constant (T), the corresponding parameters of the proposed controller are Kp = 1.5, Ti =550, Td = 0.
||The experimental results of the traditional PID controller under
constant loading of 1000W
||The experimental results of the proposed controller
under constant loading of 1000W
Therefore, we can fulfill
the accuracy of control system merely by a simple commercial available
PID controller under the precise system identification and appropriate
design of the controller. In industry application, the controller is often
handled with PC-based and DSP-based to obtain high accuracy and high performance
control. However, it got some drawbacks such as high cost, complicated
operation and difficult to maintain. Therefore, the proposed Smith predictor
controller can fulfill high accuracy only by using commercial available
PID controllers. The proposed PID control method has the advantage of
simple construction, low cost and easy regulation. It could meet the industrial
application demand extremely.
The experimental results of the temperature response under step load
change abruptly for the traditional PID control and the proposed control
are shown in Fig. 9 and 10, respectively.
From the comparison of temperature response, we can observe the experimental
results roughly resemble the simulation results in spite of the nonlinear
time varying characteristic of the cooler systems (Fig.
9). The control accuracy of the proposed control scheme is better
than that of traditional PID controller and can achieve the precise temperature
control within ± 0.3 ° C at the steady state. Besides, to observe
the effect of load disturbance, the same PID controller is compared with
the proposed controller when the loading is altered from 1000 W to 700
W abruptly at t = 1000 sec. The loading of 1000 W were added to at t =
1400 sec thereafter. From the results shown in Fig. 10,
the proposed PID control scheme still can achieve satisfactory control
accuracy of temperature for the system at the steady state.
||Temperature response for the traditional PID controller
under step load altered abruptly
||Temperature response for the proposed controller under
step load altered abruptly
However, during the step
load change period, the temperature variation becomes too large to be
acceptable for traditional PID controller.
The machine tool coolers are also the best managers of coolant temperature
in avoiding the deviation of spindle centerline for high speed machining.
However, there exists the problem of the nonlinear time varying and time
delay characteristic of the cooler system for high speed machining. In
this study, a new PID scheme for method temperature control is proposed
specific for machine tool cooler. The satisfactory set-point tracking
and disturbance rejection performance has been achieved by the proposed
control scheme. A simple commercial available PID controller based on
the Smith predictive method is adopted as a basic controller. Besides,
to deal with the nonlinear time varying and time delay characteristic
of the plant, the PWM technique is then chosen as an auxiliary duty ratio
control to modulate the hot-gas by-pass solenoid valve to provide precise
oil temperature control. Some simulation and experimental results are
also given and compared extensively for demonstrating the effectiveness
of the proposed control scheme. The control scheme not only improves the
performance of PID controllers cost-effectively but also maintains a permanent
temperature control specific for machine tool coolers.
The authors would like to express their great appreciation to the
financial support by the National Science Council under the grant No.
NSC-93-2622-E-167-001-CC3. The financial support from the Industrial Technology
Research Institute is also very much appreciated.