A high-rate filter pilot study at 9 gpm/sf was conducted at the Lanier Filter Plant in Gwinnett County, GA during the Fall of the year 2000. The Lanier Filter Plant is a 150 mgd direct filtration plant that treats water from the Lake Sidney Lanier. The treatment train consists of pre-ozonation for 0.5 log Giardia inactivation, rapid mix with ferric chloride and cationic polymer, three stage tapered flocculation and deep bed filtration. The filters consist of 12 inches of sand and 48 inches of anthracite and are variable controlled constant rate filters. The pilot study was conducted to temporarily increase plant capacity to meet the peak summer demands until a new 75-mgd direct filtration plant is constructed. During the pilot study, the pilot filter consistently ended its filter run on headloss. The terminal headloss for the pilot filter was 4.5 feet at 20 OC. However, according to hydraulic calculations the terminal headloss should have been 5.9 feet, which would result in longer filter runs; hence, greater plant capacity. The net plant capacity (12 filters in-service) with the existing 4.5 feet of terminal headloss is 164 mgd, whereas the net plant capacity with 5.9 feet of filtering head is 171 mgd. The missing 1.5 feet of filtering head corresponds to an extra 5 mgd of plant capacity. To find the missing headloss, field measurements were made through the filter and effluent piping and then compared with calculated values. Next, pressure measurements were taken along the filter effluent piping at different flow rates and compared to calculated values. Next, pressure measurements were taken along the filter effluent piping at different flow rates and compared to calculated values. Pressure measurements were taken on the blind flange of the tee, at the venturi entrance, at the venturi throat and on top of the 90 degrees elbow. The measured headloss from the tee to the venturi throat matched the calculated headloss very well. However, measured and calculated headloss values between the venturi throat and elbow differed by more than 2 feet. The difference between the measured and calculated values decreases with flow rate. To determine if this is an isolated occurrence to Filter 12, pressure readings were taken at Filter 3. Again, excessive headloss across the flow control valve was observed. When the venturi manufacturers were contacted, they mentioned the extra headloss could be from the hydraulic interaction of the venturi and the butterfly valve. To verify the hydraulic interaction between the venturi and butterfly valve, the filter effluent piping was modeled using 3-D computation fluid dynamics software Fluent 5.5 Includes table, figures.