Controlling the prefeeder

Controlling the prefeeder A prefeeder is a modulating device of some kind, controlling the material flow into the belt feeder infeed. You need it for three reasons: 1. The material can not be sheared at the belt infeed because it (at least sometimes) behaves like water. 2. The material in the silo above produces such a vertical force that the belt feeder can t handle it. 3. There is no mass flow path from the silo to the feeder. Typically, this would be an air slide arrangement. Prefd Delay PreFeeding Device Infeed LdCell Length Load Cell So now we have two devices to control, the belt (speed) and the prefeeder (feedrate). Normal belt speed control is easy, because the action-reaction time is predictable and does not change with process parameters. It is made up of belt speed inertia and ramp limitations on the belt drive system. Prefeeder control is nasty, because the actionreaction time is hard to predict, and depends among other things on the belt speed, which typically varies. In reality, over long enough time, the prefeeder determines the feedrate, not the belt speed. Speed Cascade Control It is fairly common to let the output to the belt speed drive system also control the prefeeder drive system. This means that the prefeeder is the ultimate controlling device, and the belt speed is adjusted so that reasonable belt loading is maintained. With a Position Adjust Type (PAT) prefeeder this solution will typically not work, because Stopping the belt and closing the PAT simultaneously will most likely cause feeder flooding, because the belt stops before the PAT has closed completely. The max feedrate through the PAT may be higher than the belt feeder capacity. This may cause feeder flooding at high setpoints. There is a flooding hazard when the setpoint is reduced quickly, since the belt may slow down quicker than the PAT reduces the feedrate. The PID tuning is very difficult, because the belt speed will be one component of the process delay, meaning that the process parameters are setpoint dependant. There is no PAT controller available, that that takes a 4-20 ma input. The MC³ 20.20.EX.D controller firmware provides for handling this scenario. There are numerous filters and parameters to optimize the control algorithms. The PID parameters have to be tuned for the longest delay from the prefeeder to the weigh suspension. To do this, you have to run the feeder at minimum (minimum belt speed, max delay) and then tune the PID. This is not an optimal solution, but it may be the only one available. Copyright 2005-2009 Merrick Industries 1 of 7 09/10/09/Lars

Speed Cascade Prefeeder Control The numbers in this figure refer to the actual MC³ register numbers. The variables are accessible for logging and trending with WinMerik, a must if you want to tune the system optimally. Copyright 2005-2009 Merrick Industries 2 of 7 09/10/09/Lars

Set the Strategy (0-4) parameter to 2 or 3 in the Prefeeder Settings menu to use the Speed Cascade control mode. With a setting of 2, you control (set up) the Feed Factor. With a setting of 3, the Feed Factor is updated by the controller, using the Hi and Lo Load parameters. Before attempting to optimize control dynamics, all calibration procedures have to be completed. Don t forget the position signal from the PAT position transmitter. PAT manufacturers try to make their valves follow a linear relation between position and feedrate. This effort can be rendered useless in the PAT position value is not properly calibrated. The PAT position is available in R[221]. It is also necessary to measure the time it takes for the PAT to go from fully close to fully opened and then enter this value in PAT Stroke Tm. Many PAT s have sticktion and hysteresis. This will generate a wobbling feedrate. It is possible to remedy this situation somewhat by setting the Dead Band parameter low, thereby causing a slight oscillation in the PAT position value. The Feed Factor parameter must be set by hand in Manual Speed Cascade mode (2). The starting point is 100%, assuming that the PAT will generate 100% design feedrate when fully open. Since most PAT s have some over-capacity, it will have to be reduced, so that you have a reasonable belt load at full setpoint. I Automatic Speed Cascade Mode (3), the controller will adjust the Feed Factor such that the Load is maintained. The Filter Speed IIR (Speed parameter in the Dampening Factors menu) should be set so that speed value jitter is neglectable at the lowest speed you will use. Note that this filter is inside the PID loop. You may have to re-tune the PID if you change it. The Filters Load IIR and Load FIR (Load Damp and Load Slots in the Dampening Factors menu) should be used to low-pass filter the Load variable. Set Load Damp to 0, and then increase Load Slots to eliminate short term noise on the load signal. Note that this filter also is inside the PID loop in Speed Cascade mode. You may have to re-tune the PID if you change it. The Filter FIR ( in the Dampening Factors menu) affects feedrate indication and the associated analog output only. It plays no part in the internal MC³ control dynamics, but may play a part in outer control loops. There is a clear tendency to apply too much feedrate indication dampening, because it makes the feedrate trend record look good. Note that a good looking feedrate indication not necessarily means a good feedrate. The Filter FIR ( in the Dampening Factors menu) is used to low-pass filter the user s setpoint value. Just like the Filter IIR, it plays no part in the internal MC³ control dynamics, but may play a part in outer control loops. The obvious use is to eliminate noise from 4..20 ma signaling, but it can also be used to avoid flooding of the transport system at feeder start and of feeder flooding when the setpoint is drastically reduced. The Ramp Limiter (SCR Accel and SCR Decel in PID Settings) is in place to protect the drive and drive train. Most modern drives have better built-in protection. If you have problems with feeder flooding when the setpoint is drastically reduced, you can decrease SCR Decel. Be aware that it introduces a non-linear behavior inside the PID loop. PID controller tuning is typically up the controls engineer s experience, but Ziegler Nichols method is usable in a setting like this. The tuning should be done at the lowest setpoint you will use in normal operation, and after ramp limiters and speed and load filters have been set up. At full setpoint, the controls may appear sluggish. Don t fall for the temptation of increasing the gain. It may cause the system to oscillate at low setpoints. Prefeeder Based Step-Change control Looking at the prefeeder alone, a much better solution is to us a step-change type control, with variable process delay. A reasonable assumption is that there is a fixed delay from the prefeeder to the belt infeed, and then a belt speed dependant delay from the infeed to the weigh suspension. This type of control is easy to tune. You need the prefeeder-infeed delay time and Copyright 2005-2009 Merrick Industries 3 of 7 09/10/09/Lars

the prefeeder reaction time. The feeder controller knows the delay between the infeed and the weigh suspension. Load / Speed Delay Belt Motor PID Prefeeder With this arrangement, what do we do with the belt speed? Why not run the belt in speed control in such a way that you get a reasonable belt load at all times. The desired belt speed can then be calculated easily out of the desired belt load at min feedrate and the desired belt load at max feedrate. Step Change Alternate Load-Based control Step-Change control It is possible to run the belt speed control algorithm based of feedrate (as opposed to a calculated speed ). The basic step-change prefeeder control is still in place, but the prefeeder output in nudged so that a desired belt load is reached. This control mode is more difficult to tune, but in beneficial under two circumstances: Speed Delay Belt Motor PID Step Change Prefeeder Nudge 1. The prefeeder produces a material flow that varies short term. 2. The feedrate is mostly constant. An erratic or pulsating prefeeder feedrate can then be evened out by varying the belt speed. Long term, however, the prefeeder still determines the feedrate. Load Copyright 2005-2009 Merrick Industries 4 of 7 09/10/09/Lars

Using a level sensor in the infeed chamber If there is a level switch in the infeed chamber, it can be used to control the prefeeder. The control scheme is simple increase the prefeeder demand signal when the switch is disengaged (no material at switch) otherwise decrease it. Use Strategy 4, connect the level switch to an unused input, map the input to the Mat on Belt logical input. The Feed Factor parameter is increased or decreased with the value of Max FF Step on an interval determined by Transport Dly. All parameters are available in the Prefeeder screen. CAT and PAT prefeeder actuation The prefeeder actuator is often some kind of modulating valve. It requires two outputs, Increase and Decrease, and normally produces a return signal indicating the position of the valve. This is called a PAT (Position Adjust Type) device, as opposed to s CAT (Current Adjust Type) device. A typical CAT device would be a rotary feeder that is driven by a variable speed device, following a demand signal. A Blind PAT is a PAT device without the return signal. Blind PAT control is possible (sometimes) but not recommended. Don t flood the belt! If there is a significant delay between the prefeeder and the infeed, you run the risk of flooding the belt when the feedrate is drastically reduced. The prefeeder control signal and the belt speed are both reduced, but for a while, the feedrate reaching the belt will remain at the previous, high. The same can happed if the feeder is unconditionally stopped with the feeder block input or with an emergency stop arrangement. Both the belt and the prefeeder will stop, but there could still be a considerable amount of material in the pipeline. There are two ways of preventing this to happen: 1. Apply filtering 2. Reduce the permitted rate of change to the belt speed output A failing belt drive system or a slipping belt will also likely flood the feeder. When the feeder is stopped, either by removing the Run Permission input or hitting the Stop button in the Feeder Control screens, the belt keeps running for a while. The delay time is the sum o the PAT Stroke Time and Predf Delay. See MC³ prefeeder parameters, below. If you are using a PAT device, it is a good idea to reverse the logic of the PAT Adjust Down logical output, so that a closed relay output means don t adjust the PAT down. This will prevent the PAT to be left open if power is lost to the MC³ controller. MC³ Prefeeder parameters When you set the logical input Prefeeder to Always On in the MC³ 20.00.EX.D, an entire system of parameters is enabled. For successful operation, all of them have to be set properly. The concept of a load appears whenever you have a prefeeder arrangement. Other parameters are the dead time from the prefeeder to the infeed, the prefeeder s reaction time, and, since we are dealing with a step-change control algorithm, the max allowed change in the step. If you use a PAT device, the PAT reaction time (Fully Open to Fully Closed) has to be measured. The MC³ controller makes a linear assumption of the prefeeer characteristics, maintained in the variable Prefeeder Factor, in percent. This variable is simply used to set the output to the prefeeder drive, by multipliing it with the feedrate. It is then updated on a cyclic fashion, based on the total delay from the prefeeder to the weigh suspension. The primary benefit is that the reaction time will always be optimized to the current belt speed. The faster the belt speed, the sooner the feedrate will come to the. Another great benefit is the fact that the prefeeder will react immediately to a change. Copyright 2005-2009 Merrick Industries 5 of 7 09/10/09/Lars

In the MC³ 20.10.EX.D application, the following parameters have to be considered. In the Prefeeder Params menu: Max and Min Load SP. The desired load at max and min, found in the Limit Switch menu. You can set both of them to the same value for a constant load operation. It is also possible to set them proportionally to the max and min, producing a constant belt speed operation. Reasonable settings are 90% and 50% of the design belt load, respectively. Ctl Dead Band is used to leave the Feed Factor alone when the belt load is within the Dead Band range of the load setpoint. A reasonable value is 5%. PAT Dead Band is used to leave the PAT alone when the position signal is within the dead band of the nominal position. A reasonable value is 2%. If you are using a PAT, check the PAT UP and PAT DN indicators in the main screen, and adjust the dead band to reasonable operation. You don t want to wear out the PAT by constantly driving it up and down. Start Delay. This is a hold-off time for the prefeed controller update at feeder start. The default is 20s. Some arrangements require a long time to become operational at feeder start, so this time may have to be adjusted. Transport Dly. This is the material flow delay time between the prefeeder device and the belt. For a rotary feeder just above the infeed, set a value of 0. The parameter is critical, as it affects the response time. If there is some kind of transport going on between the prefeeder and the belt, this time has to be considered carefully. Using strategy 4, this parameter determines how often the Feed Factor is updated based in the status of the Mat on Belt input. PAT Stroke Tm. If you are using a PAT, this parameter must be set to the PAT reaction time, that is, the time required to go from fully closed to fully opened. Feed Factor. This is the Prefeeer Factor, controlled by the step-change algorithm. You can override it in the menu. Under ideal conditions, it should hover around 100% Max FF Step. Max allowed change in the Prefeeder Factor per step. A reasonable value is 1%. The higher the value, the more aggressive the control. FF Pullback. Used in strategy 3, in combination with the load limit switch values (Limits menu), to drastically change the Feed Factor when the low and high load limits trigger. This parameter determines how much the Feed Factor changes. 10% is reasonable. AIN Lo and Hi Counts. These are calibration constants for the PAT position analog signal. Default is 200,000 and 1,000,000, respectively. The A/D converter used has a range of 0 1,048,575 counts. The parameters are only used if you are running a PAT with an analog position return signal. If you have done an Analog Input Calibration Procedure for the analog input for the PAT return signal, the values will change accordingly. AIN Zero and Full Val. Used to skew the scaling of the PAT position analog signal. Default is 0 and 100%. There is normally no reason to change these values. Strategy 0-4. Control mode selector. Default is 1, -based mode. 0, Alternate Load-Based control Step-Change control, page 4. Make sure that the feeder normally operates with a constant. 1, Prefeeder Based Step-Change control, page 3.Step change control of the prefeeder, speed control of the belt, in such a way that the desired belt load is maintained at feedrate setpoint. 2, Speed Cascade Control, page 1. Feed Factor is set by the user Copyright 2005-2009 Merrick Industries 6 of 7 09/10/09/Lars

3, Automatic Speed Cascade Control, page 1. Feed Factor is updated automatically. 4, Using a level sensor in the infeed chamber, page 5. In the Analog Inputs menu: The Prefeeder Button has to be set for what kind of prefeeder you are using. Set to Always Zero for CAT or Blind PAT. Set to the appropriate Analog Input for a PAT with an analog position signal. Set to Remote for a PAT with a networked position signal. Copyright 2005-2009 Merrick Industries 7 of 7 09/10/09/Lars