Monitoring Water Contamination

Mosaic is the merger of two global leaders in the fertilizer industry, IMC Global and Cargill Crop Nutrition. As one of the world's largest potash and phosphate mining and processing operations, Mosaic provides an expanding selection of products and services to enhance crop yield and livestock nutrition.

Figure 1. Test apparatus used for analysis

One of Mosaic's potash mining sites, the Belle Plaine plant in Regina, Saskatchewan, relies on two General Electric 20 MWe turbine generator sets for co-generated power and process steam, and three Westinghouse EL125 prime movers for process applications. A major consideration for any turbine operation is the development of a comprehensive preventive maintenance program to ensure long service life. A major destructive contaminant to a turbine's bearings and hydraulic system is water. Despite the measures in place to prevent water ingression into a turbine's lubrication system, water contamination is an inevitable phenomenon that should be carefully managed. Sources of water ingression include the high pressure steam driving the turbine, condensation of water vapor in the head space of a lubrication reservoir and failed bearing seals.

Figure 2. HMP228 transmitter installed on #4 turbine engine

Due to continuous moisture problems, the Mosaic technical services group initiated a research project in 2005 to actively identify moisture levels in the turbine's oil and hydraulic systems. To accomplish this goal, an online oil sensor to measure water concentration needed to be chosen and calibrated to their needs.

Figure 3. HMP228 probe is inserted directly into the line with ball valve assembly. The unit is installed in a bearing return oil drain adjacent to the high-pressure end turbine packing.

Mosaic chose Vaisala's HMP228 moisture in oil instrument to provide in-line, continuous measurement. The instrument displays the reading and provides an analog output signal and adjustable alarm relays. The HMP228 directly measures water activity (aw), which indicates a fluid's margin to saturation on a scale of zero to one; zero being completely dry and one being fully saturated. Before installation, Mosaic needed to characterize the performance of the instrument as a function of temperature and water concentration (in ppm(mass)) as a variable of aw output. A test apparatus was created consisting of a steel drum placed on a heating plate with an agitator to circulate the oil. The apparatus was created to evaluate how the sensor responsed to specific conditions and to calibrate the sensor to Mosaic's specific needs.

Three different test procedures were designed to evaluate the sensor performance.

Test Objectives

• Test 1.Determine if the HMP228 transmitter will respond to the oils used in Mosaic's turbines at the Belle Plaine location.

• Test 2.Determine how temperature affects the water saturation level of the virgin turbine oil.

• Test 3.Determine how the HMP228 transmitter responds to water content changes in turbine oil.

Testing
Test 1
The two oils used for the test were Petro-Canada TURBOFLO Turbine Oil (virgin oil) and Shell VSI Circulating Oil 32 (water saturated oil). Each oil was placed in the test apparatus and heated to 115°F. The HMP228 probe was inserted into the oils and allowed to stabilize. The virgin oil measured a water activity (aw) of 0.091 with a visual observation noted as "clear, transparent." The water saturated oil read an aw of 0.968 with a noted appearance of "foggy, cloudy, unable to see bottom of drum", according to the Mosaic technical services group.

Test 2
Using the test apparatus, the turbine oil was heated from room temperature to 140°F while water activity readings were recorded with the HMP228. The data was plotted and a linear trendline was created.

As expected, the water activity readings decreased as the oil temperature increased.

Test 3
Maintaining the turbine oil temperature at 115°F, known ppm volumes of water were added to the oil and water activity readings were recorded at each point. The data was plotted and a linear trendline was created.

As anticipated, the water activity readings increased linearly as the water content of the oil increased.

Results
A regression analysis was completed for both oils and a contour plot was developed to characterize water activity as a function of both temperature and water content.

With the good results from the testing, Mosaic installed the HMP228 on both 20 MWe turbine generators. "Based on this positive result, our intent is to alarm at 0.60 water activity. Both temperature and water activity will be used to calculate ppm water in oil within plant DCS systems. The ppm water calculation can be compared to oil analysis results. Corrective actions will be taken beyond 0.65 aw and can include external scrubbing, corrections to gland condenser systems, and packing inspections," said a member of the Mosaic technical services group.

Reaping the Benefits
Due to the installation of the HMP228 with the alarm relay function activated, Mosaic was alerted to a potentially destructive water ingression problem caused by water entering the oil system through a faulty bearing packing that was spilling steam into an adjacent bearing labyrinth seal.

The installation, calibration and field testing of the HMP228 is a typical example of collaboration between functional groups within Mosaic and vendors who support Mosaic's objectives with technology for process improvement.

Author's Note:

Special thanks goes to the Mosaic Technical Services Group for their technical expertise, cooperation and sharing of results and performance test data for this research study.