Sheath temperature monitoring for rotary kilns
Initial situation and project task

Humboldt Wedag GmbH (KHD) is one of the world’s leading manufacturers of industrial equipment for the cement industry. In addition to rotary kilns and their control modules, KHD also provides systems for temperature monitoring, process control and quality management, called SCANEX. These systems consist of infrared detector heads which are installed opposite to a rotary kiln at a defined distance. Within their angle of view these heads detect the temperature profile of a kiln’s sheath which is transmitted to the control room for visualization.

Scanex Single Board RIO NOFFZ Technologies - Produktentwicklung und Serienfertigung

Based on that temperature profile decisions on future filling, heating and rotational speed of the kiln can be made. Besides full control of the clinkering process itself, SCANEX allows early recognition of wear and damage to the inner brickwork of the kiln, so that expensive damage to the kiln can be avoided. Due to the steadily growing demand for more measurement data precision and processing velocity, KHD had to increase data resolutions as well as data acquisition rates of their detector heads. In order to increase processing velocity for cement production by meeting the steadily growing demands, maximum rotational speed of kilns has been augmented from 3 U/min. to 6 U/min. In addition to that, the bit resolution of digitized measurement values had to be increased. The data acquisition rate of currently 21.6 kHz should be augmented to 72 kHz.

During warm-up at process start, rotary kilns can have very long rotational periods of up to 12 minutes. Measurement data is collected and buffered for the whole rotational cycle so that an additional signal is required and evaluated which indicates beginning and end of each rotation. The frequency of the rotational signal ranges from 1 mHz to 100 mHz; in comparison to its duration its pulse width is rather small (< 1%).

Implementation & challenge

The main challenge presented during the implementation of this system was the large spectral range of signals which had to be acquired and evaluated. The smallest frequency to acquire was even smaller than 1 Hz while the fasted signal to detect had a frequency of about 72 kHz: The hardware solution had to be able to acquire an analog value from two detector heads, digitize both values at 12 Bit and transfer them to a buffering unit at high speed, triggered by a high-frequency signal. In addition to that high-speed triggered action, the system had to be able to react to the low-frequency rotational signal, which starts the evaluation of the whole temperature profile.

The system had to provide sufficient storage capacity for buffering large amounts of data which can be gathered during a cycle. Operating at a data rate of 72 kHz and at a minimum rotational speed of 12 minutes, several Mega Bytes of data have to be stored despite all compression methods for redundant information.

On the one hand, a suitable hardware module for this task has to offer a sufficient aggregate sample rate for all channels in order to digitize signals at a rate of 72 kHz. On the other hand, the module has to provide independently triggerable inputs, so that signal frequencies between 1 mHz and several kHz can be evaluated separately and in parallel. In order to realize independent triggering for a wide frequency range, an FPGA module by National Instruments has been chosen. The choice of a sbRIO9633 guaranteed for short development times for the bearing board for the FPGA system. This board provides analogue and digital inputs of the FPGA board with common connectors so that further signal conditioning is redundant.

In addition to the FPGA module the sbRIO board provides an RT module which is capable of processing measurement data within the system-defined period of time. It can process and forward data to the host PC/control room, deterministically. For these reasons, the sbRIO 9633 has become the hardware core of the monitoring system: It acquires analog values from up to two detector heads, converts them into 12-bit digital signals and forwards them to the RT system. This procedure is triggered by two independent digital inputs, whose signals display highly different frequencies and pulse widths.

All software modules for controlling data acquisition and processing have been implemented using LabVIEW2011 SP1. The FPGA integrated into the sbRIO 9633 reliably and smoothly fulfills its task of analog and digital data acquisition with extreme different data rates. The integrated RT system performs real time data processing even of large data amounts provided for long rotational periods. After processing, it forwards data to the host PC, in time. The host conducts graphical data preparation and visualizes the temperature profile on the kiln’s sheath offering different views.

Conclusion

Based on a sbRIO9633, a robust, fast and reliable data acquisition system for temperature monitoring of kilns has been designed and implemented. Because of its integrated FPGA module, the system is capable of digitizing measurement values at a high resolution, in time, and of acquiring analog and digital signals whose frequencies range from less than 1 Hz to several kHz, in parallel. The RT system is able to analyze and process measurement data as fast as required, so that graphical presentation and data storage can be conducted by a Windows host system being less optimized for velocity.

The demand for rotary kilns steadily increases all over the world, especially in Russia and India. Therefore, the new SCANEX will soon be found at cement plants all over the world. In order to guarantee optimum integration into each plant, SCANEX will be equipped with numerous optional modules, allowing control rooms accessing measurement data via interfaces such as OPC.