Leaks pose ongoing process challenges. Plants with heat exchangers, condensers or jacketed vessels face the prospect of the cooling water or other heat-transfer medium leaking into the process or vice versa. This can cause contamination and loss of product. Leakage can prompt even larger problems when it leads to destruction of valuable equipment, as the FujiFilm chemical plant in Dayton, Tenn., can attest.
The site, which makes photosensitive chemicals, feeds dry materials and flammable solvents, including thionyl chloride (SOCl2), to a reactor used to precipitate crystals. The volatile solvent is pulled from the reactor and routed through a heat exchanger using a liquid ring vacuum pump. Unfortunately, SOCl2 produces hydrochloric acid when it comes into contact with moisture. When acid from the process got past the pump's seal, it damaged the pump. In the last three years, the $28,000 pump was replaced multiple times due to corrosion.
Three alternatives could prevent the problem: using exotic materials of construction, installing an intermediate tank to capture vapor (not possible in the available space), or analyzing the seal fluid to detect leaks before they caused damage. Opting for a simple analysis system was the proverbial "no-brainer."
Leak detection can be performed using either pH or conductivity analysis. The choice depends on the process. For pH to be used, a small amount of contaminant must cause a measurable change in the pH of the process; for conductivity to be suitable, the contaminant must significantly alter the conductivity. Conductivity can detect leaks of acids, bases or even salts but requires stable process conductivity for best results.
The chemicals to be monitored by FujiFilm affected pH. So, the plant installed a pH analyzer in the vacuum pump seal loop (Figure 1). It chose a wireless unit to obviate power and output wiring. Because a wireless gateway already was in place for other process control applications, implementing the analyzer was easy. It was incorporated into a self-organizing network that allows each device to function as a data repeater. Thus, if any pathway becomes interrupted, data automatically travels via an alternative pathway, assuring uptime. The pH monitoring system cost less than $3,500 to implement and was up and running in two days.
Since the plant installed the pH analyzer in June 2011, it hasn't suffered any corrosion-related pump failures.
In the FujiFilm application, the normal pH of the seal fluid (water) is approximately 7; at that point, corrosion is minimized. To protect the vacuum pump seal integrity, when the analyzer finds the pH has dropped below 3, the process is stopped and the system is flushed to clear out the acid and return the process pH to 7.
The FujiFilm application is an excellent example of appropriate use of pH-based leak detection because when the process pH is near neutral (7.0) the pH response is greater, resulting in increased sensitivity to detect leaks. In general, however, pH may respond differently to the presence of contaminant, depending upon the process pH and the chemistry involved.
In dilute processes or ones without buffering action, it's easy to predict the pH response. Table 1 shows the best leak-detection sensitivity possible for various process and contaminant pH values. It gives sensitivities in terms of the volume fraction of the contaminant detectable, expressed from percent through parts per billion by volume. Estimating sensitivity requires considering the normal pH range of the process and the contaminant, checking the maximum and minimum pH of each against one another, and then using the poorest sensitivity.
Processes containing weak acids and bases, such as acetic acid, hydrofluoric acid, ammonia and sodium carbonate, have weaker pH response, which reduces the sensitivity of leak detection. This is due to buffering action, i.e., the tendency of a solution to resist pH changes. It often is difficult to predict just how the process pH will respond in such mixtures. Titrating a sample of the process with the contaminant will give the sensitivity — it is the volume of contaminant required to cause a ±1.0 pH or other reliably measured pH change divided by the volume of process fluid titrated. As an initial check, consult Table 1 to see if there's the possibility of acceptable sensitivity before titration.
For FujiFilm, the implementation of pH analysis for leak detection was a low-cost solution. The option to employ wireless technology further reduced costs and implementation time.
DAVE JOSEPH is Irvine, Calif.-based senior industry manager with Emerson Process Management, Rosemount Analytical. E-mail him at [email protected].
