No matter what process variable to be measured, the ultimate purpose of implementing on-line process analysis is to improve performance and increase plant profitability. Real-time analytical information generated about a process with these systems can be used to improve yield, better manage process performance, and limit off-spec losses. To accomplish this profitability goal, the analyzer needs to return significantly greater value than the cost to develop the application, install the system, use it daily, and maintain it.

FT-IR (Fourier Transform-Infrared) and FT-NIR (Fourier Transform-Near Infrared) are among the most efficient, fastest, and reliable process monitoring techniques available today. Having crossed over from laboratory applications, FT-IR, a proven spectroscopic technique, has now been used for process in-line, at-line, and on-line analysis for over 15 years. In some cases, the system's archived data can also be used to meet environmental regulations. Even in harsh environments, FT-IR systems commonly register up to two years mean time to the first failure and typically offer 99.5% avail-ability. Many users report payback periods of a few months.

The most important difference between process FT infrared and FT near-infrared monitors and other types of IR systems is the analytical process. The Fourier Transform infrared spectrometer in an FT-IR monitor has been proven to be extremely accurate and remain stable for years even in harsh process environments.

An FT-IR system generates measurements based on a Fourier transform. Simply stated, this mathematical process decomposes a signal into a set of sinusoidal components. In an FT-IR spectrometer, this process is performed optomechanically by an interferometer. In its basic form, shown in the FT-IR diagram, the process starts when radiation from a source is directed to a partially transmitting, partially reflecting beam splitter. Half of the beam is reflected to a fixed mirror and the other half is transmitted to a moving mirror. Each mirror then reflects light to the beam splitter, where part of each beam returns to the source. What remains of the two beams combine, and as they combine, they interfere. At the point where the beams are exactly identical in length, all frequencies of light combine constructively. As the mirror moves, each frequency goes through cycles of constructive and destructive interference.

The frequency of these modulations is proportional to the frequency of light and to the speed of motion of the mirror. The interferogram a series of sinusoids and their sum. This sum is called an interferogram and is the signal received by the detector. The interferometer therefore encodes each optical frequency with a corresponding Fourier modulation frequency. The Fourier-encoded infrared signal is digitized at the detector. Finally, the digitized interferogram is inversely transformed by a computer into a spectrum.

Each mirror scan takes approximately a second and produces an entire spectrum. Scans are usually co-added to improve sensitivity, and in a sample measurement the process is repeated twice. First, a set of scans is made with no sample to develop an instrument function. Later, a new set of scans is collected with each sample present. These scans are ratioed against the instrument function to obtain the transmittance spectrum of the sample. This way of obtaining an infrared spectrum is precise, reliable, and cost effective. Other techniques also are available.

A single-process FT-IR analyzer can generate a reading every 10 sec to one minute, depending on the application, while monitoring multiple properties. Measurements can be made on up to 35 streams. Real-time measurements from FT-IR systems can be integrated directly into a plant's control system for automatic monitoring of liquid, solid, or gas-based processes. Several application-specific systems are now offered by various vendors that can be integrated into a larger process control system.

Overall, FT-IR and FT-NIR systems offer several advantages to on-line monitoring, including:

• Fast response time (10-60 sec);
• Multiple component monitoring of different properties;
• Multiple sample-point analyses;
• One-time calibration;
• High reliability, and
• Low cost of ownership.

A system that offers a quick payback and precise monitoring results can generally be configured from standardized components. To ensure best results, control engineers should work closely with a qualified vendor to develop a system most appropriate for their application.

For more information on Analect Instruments, visit www.controleng.com/info