Filling material level measurement.

The “classical approach” to measuring the filling level

With ball mills this working point is linked directly to the level of the material being ground in the mill. A mill that is too empty, i.e. one that has been underfed, operates extremely uneconomically from the energy point of view while a mill that is too full, i.e. has been overfed, also grinds very ineffectively as the grinding balls fall on a “soft” bed of mill feed and lessen the progress of comminution along the grinding path.
In order to be able to optimize a mill it is therefore necessary to measure the level of material being ground. Direct measurement in the mill is not possible so in most cases microphones, so-called “electric ears”, were used for this purpose. They enable a (very) rough estimate to be made of the level of the material being ground in the mill on the basis of the sound intensity – experience shows that an empty mill produces a loud, high pitched, sound while a full mill produces a somewhat duller and quieter sound.

Trong các trường hợp đo mức đặc biệt như chuyển động quay, môi trường có sự va đập… thì việc kiểm soát mức của vật liệu cần phải kiểm soát bằng âm thanh hay bằng xung âm thanh. Điển hình là đo mức than và bi nghiền bên trong máy nghiền than.

Với kiểm truyền thống trước đây, một thiết bị thu âm thanh được lắp đặt hay còn gọi là tai điện “electric ears” và ước lượng mức độ đầy của vật liệu bên trong máy. Nếu ít than bên trong âm thanh sẽ đanh và to hơn, ngược lại nếu đầy thì âm thanh sẽ trở nên hỗn độn và nhỏ hơn.

The “classical approach” to measuring the filling level

The “classical approach” to measuring the filling level

 

Structure-borne sound

The problems of the ”classical approach” with a microphone can be avoided if the structure-borne sound is measured directly at the mill shell instead of measuring the airborne sound. A piezoceramic sound sensor is mounted directly on the mill shell, which solves several problems at once:

  • The structure-borne sound signal still carries all the information that is not passed on when emitted to the air, so the measurement is substantially more accurate.
  • The sensor cannot measure airborne sound, so interference noise from other mills or units does not matter.
  • The sensor attached rigidly to the mill housing not only avoids any subsequent misalignment but is also not affected by dust, so there is no need for regular cleaning.
  • The structure-borne sound in the mill shell tends to propagate more in the radial rather than the axial direction, with the result that there is no problem with measuring the first and second grinding chambers in the mill separately.

Hiện nay thì cách tiếp cận kiểu cổ điển ở trên không còn được dùng, mà cách đo mức trực tiếp bằng âm thanh do cấu trúc vật liệu phát ra “structure-borne sound” thay vì đo âm thanh thông thường theo đường không khí. Một cảm biến âm thanh kiểu áp điện (A piezoceramic sound sensor – Thường là thạch anh – Các bạn có thể tìm hiểu thêm về hiện tượng áp điện ở đây) được lắp trên vỏ máy nghiền sẽ giải quyết được các vấn đề sau:

  • Cảm biến âm thanh phát ra từ cấu trúc sẽ mang phản ánh đầy đủ thông tin âm thanh thực, không phải ước lượng như nghe qua không gian, phép đo sẽ chính xác hơn
  • Không bị nhiễu bởi các âm thanh bên ngoài hay các máy khác
  • Cảm biến được lắp đặt cố định trong gian máy nghiền có thể tránh được sự xê dịch, tránh được bụi và không cần vệ sinh nhiều.
  • Âm thanh từ cảm biến sẽ lan tỏa theo dạng sóng cầu trực tiếp thay vì đi theo 1 hướng nên sẽ không bị ảnh hưởng bởi các thùng nghiền khác nhau.

The signals that have been received are then amplified by an electronic system, which is also attached to the mill shell, and transmitted by a microprocessor via a high quality wireless link to a small receiver located a few metres from the mill. The entire electronic system and the transmitter on the mill are supplied with electricity by a cradle dynamometer.
The cradle is mounted in the housing of the electronic system on the mill so that it can rotate parallel to the mill axis; it drives a small generator as soon as the mill rotates.

Tính hiệu âm thanh qua một bộ khuếch đại gắn trên vỏ máy và chuyển đổi thành phát ra dạng sóng Wifi, một thiết bị nhận tín hiệu được lắp đặt cách máy nghiền vài mét.

Figure 2 Structure Borne Sould

Figure 2 Structure Borne Sould

The filling level of the material being ground can already be measured just by simple evaluation of the structure-borne sound intensity (Fig. 3):

Figure 3 Raw signal from a structure-borne sound sensor

Figure 3 Raw signal from a structure-borne sound sensor

Fig. 3 shows the raw signal from a structure-borne sound sensor during one revolution for an empty mill, for a mill containing a moderate amount of mill feed and for a full mill. The point of highest intensity, where the falling balls strike the wall, is readily detectable in all the signals. Analysis of the envelope curve, together with a precise angle measurement, also makes it possible to provide accurate information about the location of the bottom end of the ball charge.

Non-linear evaluation of the pulse signals

The mill used in this example was first operated empty and then full. It is clear that information about the level of mill feed in the mill can be obtained here just from the averaged plain intensity signal. The dynamics of the raw signal for the empty:full ratio is at least 5:1. However, this result is not sufficient for more accurate filling level measurements that would also be suitable for setting up a closed control loop. The question therefore arises as to whether even more precise information about the filling level can be obtained from these signals. Detailed investigations carried out on such signals showed that the “classical” methods, such as filtering out frequency bands with the aid of the Fourier transform followed by calculation of the average intensity in the frequency band, do not lead to any substantial improvement. A Fourier transform presupposes that the signal to be investigated is at least in a steady-state condition over the period covered, which does not apply to the acoustic signals from a mill.
Typical noise spectra from ball mills tend to be composed of different short time “transient” signals that result in a random distribution on the time axis. A different picture is obtained if the noise from a ball mill is investigated on the basis of the occurrence of different categories of such signals.

Figure 4 the intensities of different signal categories

Figure 4 the intensities of different signal categories

Fig 4. shows the intensities of different signal categories, plotted against time, during the period when the initially empty mill is being slowly filled while in operation. One category that correlates very well with the mill feed filling level is clearly visible in the diagram. A very precise filling level signal can be derived from this after appropriate statistical analysis and calculation of the line of best fit (yellow line in Fig. 4), see Fig. 5:

Figure 5 analysis and calculation of the signal line

Figure 5 analysis and calculation of the signal line

Tests on different mills have shown that when, for example, this system is applied to a cement mill it is possible to measure the difference in filling level corresponding to throughputs of 80 and 81 t/h.

Automatic calibration

As a rule it is always possible to find several signal categories in the structure-borne sound signals from ball mills that have statistically significant correlation with the mill feed filling level. This makes it possible to calibrate the system automatically. The ball mill is operated empty and then full for the calibration. During this period all the signal categories are stored so that the signals with the greatest significance can be determined automatically. This calibrates the system, although the filling level signal obtained is naturally a relative value rather than an absolute one as it is only possible to specify the difference between the two operating states, namely “empty” and “full”. This range is then converted to a mill feed filling level of 0 to 100 %. The quality of the calibration therefore depends on whether the mill was actually operated when empty and full respectively. There are no further requirements. However, there is the problem of having to define when a mill is actually “full”. In practice it has proved appropriate to assume that the optimum working point lies at a mill feed filling level of 100 %. However, it is possible at any time to feed the mill at an even higher level, so the display range of the measuring system is arranged to be 1 to 130 %. In addition to this it is always possible to redefine the current filling level as the new 100 % value for subsequent improvement of the calibration.

Ở chế đô calibration tự động, ở 2 chế độ empty chưa có than và thêm than và giả định một mức đầy than cho thùng nghiền, thiết bị sẽ tự động xác định mức từ 0 – 100% tương ứng.

 

Facebook Comments

Leave a Reply

Email của bạn sẽ không được hiển thị công khai. Các trường bắt buộc được đánh dấu *

*