Maik Ulmschneider

Lee Tunnel Pump Rag Test

About the Project

Thames Water Utilities Ltd is the UK’s largest water and sewerage company, serving 14 million customers across London and the Thames Valley. London’s 150-year-old Victorian network can no longer accommodate the excess sewage and rainwater. Therefore the 7.6 km Lee Tunnel is the first of two tunnels designed to capture and redirect 16 million tons of stormwater and sewage as part of the 25-km Thames Tideway Scheme. It will run beneath the River Thames through Central London.

Buried 85 m underground six single-stage end-suction vertical waste water pumps (KSB SVP- 84) will lift the stormwater and sewage back to the surface, operating in a working range between 3,050 l/s at 87 m and 1,950 l/s at 17 m.

Figure 1: Size Comparison

Rag Test Requirements

Clogging of pumps is a common problem for water companies. When fibrous material, plastics and other debris come together solid ropes can be formed which can lead to reduced pump performance and even pump failure. Then, there is the unpleasant matter of removing the blockage. For crucial pumping stations, such as the Lee Tunnel Beckton Shaft, it is essential to reduce the risk of pump clogging to an absolute minimum.

MVB, the consortium comprising Morgan Sindall, Vinci Construction Grands Projects and Bachy Soletanche, which is working with Thames Water to construct the Lee Tunnel, have paid great detail to this aspect of the Lee Tunnel project. Lengthy comprehensive ‘Rag Tests’ Maik Ulmschneider

have been performed on the SVP-84 Main Pump by its manufacturer KSB at its GIW Industries factory in Grovetown, Georgia USA.

The purpose and form of the ‘rag test’ was defined in the Main Pump specification:

 The Main Pump selection had to be subjected to a performance trial on the ability to pass soft solids and grit, with the trial consisting of a flow test. The primary objective was to determine whether the pump remains within performance curve allowable tolerances and thereby free from blockage.  The test would take place consistently over a six hour period and the pump would be operated at the minimum, intermediate and maximum flow conditions during the test.  The liquid to be pumped should be a mixture of clean water and specified volumes of non-hazardous material to simulate the rags and grit quantities normally found in sewage. The mixture should consist of 2kg/m³ of shredded cotton and polyester rags (dry weight) and 400mg/l of grit. The rags should vary in width and length (up to 1000mm x 75mm).

Figure 2: Volume Comparison for Water, Rags and Grit

Rag testing of this magnitude (concentration, duration and size) is not described by any industry standard and is not state of the art. Therefore, considerable discussions took place between the manufacturer and the client to determine an appropriate form of the test and to prepare contingencies for unexpected difficulties. This resulted in a detailed test protocol:

 The test pump would be constructed using the actual hydraulic parts of the Lee Tunnel Project KSB SVP-84 Main Pump. However, due to the restrictions of the test Maik Ulmschneider

bed they were mounted horizontally using the side plates, stuffing box and mechanical end from a standard LSA-84 pump.  The flow rate, head and speed of the pump would be limited by the available drive power and layout of the piping system.  The results of the Preliminary Clear Water Performance Test would be used as the baseline performance for comparison during the rag test.  Shredded cotton and polyester rags would be loaded to achieve a resulting mixture density of 2 kg/m³ (dry weight) rags, with the rags varying in width and length up to 1000 mm x 75 mm.  If rags appeared to be massing due to the turbulence associated with repeated passes through the pump, some adjustment of the test protocol would be required.

Figure 3: Portion of Rag Set 2

Test facility

To perform the clear water and rag tests, it was necessary to construct a purpose-designed test rig at the GIW Hydraulic Laboratory in Grovetown, USA, using the 900-DN test loop facility. The first stage was to assemble the pump, followed by the assembly of the discharge and external piping. The piping was then connected to the existing 224.5 m water tank and the intake side of the pump. Due to the nature of the rag test, it was necessary to make modifications to this tank. These included the construction of a new tank floor to channel the flow from the tank inlet to the outlet, and the installation of an upper port through which the rags could be inserted and a lower port to facilitate their removal.

On completion of the construction the system volume was estimated at 244.1 m³ requiring 488.1 kg of rags and 97.6 kg of grit. However, once rag testing began it was not practical to fill the tank completely, so the system volume was reduced to 188.6 m³, 377.1 kg of rags Maik Ulmschneider

and 74.5 kg of grit. The first of the rag tests started in September 2012, following preliminary clear water performance tests.

Figure 4: Model of Test System

Rag Test Results

During testing, the rags stuck together into tightly woven masses to a far higher degree than expected. Ropes up to 3 m in length and compact clumps approaching the pump’s free passage clearance size of 375 mm were observed. It was evident that these formed during repeated passes through the highly sheared flows at the impeller outlet and at the system throttling valve where the forces needed to tear and weave the rags together are at their strongest. It was estimated that the average rag passed through the pump and valve once every minute at the highest test flow rate of 2,845 l/s, allowing ropes and large clumps to form in less than 90 minutes.

Maik Ulmschneider

Figure 5: A 3,000 mm+ Rag Found in the System after Test 3

Because of these difficulties, the test protocol was revised to bring the actual conditions of the test back in line with the requirements of the specification. Five different tests were run with rags of varying sizes and loading conditions. As a result, much was learned about the ability of the Main Pump to handle rags both meeting and exceeding the required size at various concentrations.

During Test 1 large clumps of material were observed passing by the observation port and these clogged the pump. When removed from the suction, the many ropes and clumps present were the first indication that the problem was more serious than anticipated. From this experience, it was agreed to alter the procedure for Test 2.

For Test 2, a similar mix of rags was prepared, but sorted by size with the intention of loading the shortest rags first and moving to larger sizes as the test progressed. During the first 90 minutes, with a concentration of approximately 0.69 kg/m³ and a starting rag length of <250 mm the pump did not clog. Losses were around 3% on head and 6% on efficiency. However, once rags of up to 500 mm were added (bringing the concentration up to 1.33 kg/m³) clogging occurred and the test was terminated. Examination of the ropes and clumps showed that they were tightly and firmly twisted.

For Test 3 it was proposed that there should be a particular rag length, which if loaded at the full concentration of 2 kg/m³ and run for six hours would form rags up to 1,000 mm long. This would produce a rag test meeting the requirements of the contract specification. Therefore, a quantity of 100 mm rags was prepared and Test 3 was performed exactly in accordance with the original test protocol, but with one exception: the starting rag length. Maik Ulmschneider

All the rags were loaded as fast as possible and the standard test started once a full concentration of 2 kg/m³ was attained.

As expected, rags up to and in excess of 1,000 mm were created. Based on observations of pump head and power during the test, the formation of these ropes and lumps peaked between 90 and 120 minutes into the test after which point they started breaking down again. It was also noted that a significant amount of rag sludge and fragments were produced by the end of the test. Rags meeting and exceeding 1,000 mm remained until the end, including one that exceeded 3,000 mm. Head and efficiency remained within contract limits throughout the test, with the exception of the short period when the 3,000 mm rag was lodged in the pump suction.

Figure 6: Head vs. Time Test 3

While the results of Test 3 were being compiled, it was decided to run more tests to further quantify the effect of rags on the pump and the dynamics of rag formation during the re- circulating test. In Test 4 it was decided to repeat Test 3, but with a larger rag size of 200 mm to determine the effects. During this test, rope and clump formation proceeded more slowly than in the first two tests, but some large clumps formed within the first 60 minutes after loading commended. As these grew to a size in excess of the pump free passage, they repeatedly clogged the pump, reducing performance. The pump was able to keep clearing itself for a few minutes until the size and number of clumps became excessive.

Maik Ulmschneider

Figure 7: Some of the Rags Found in the System after Test 4

Figure 8: Head vs. Time Test 4

Apart from clogging the suction side, ropes, strands and clumps were not observed in any other areas of the pump, such as the casing, pump cutwater or behind the impeller, which are common sites for blockages and binding in conventional sewage pumps.

Test 5 was run in accordance with the client’s wishes to determine the effect of adding rags up to 1,000 mm long from the beginning, but starting with a lower concentration and increasing thereafter. At a concentration of 0.5 kg/m³ (or 25% of the required 2 kg/m³) the pump remained near tolerance on head, even though longer rags were forming. After 90 minutes of continuous operation, the rag concentration was increased further. When it exceeded 1 kg/m³ clogging resulted. As in earlier tests, rag masses far exceeding the maximum requirement of 75 mm x 1,000 mm were formed, but more slowly due to the reduced concentration at the start of the test. Maik Ulmschneider

Conclusion

The recorded data and video footage from Test 3 showed that it met or exceeded the requirements of the rag test specification. In addition, Tests 2, 4 and 5 provided information regarding the limits of the Main Pump in handling rag masses well in excess of the specification requirements.

Key learnings include:

 The Main Pump can handle a mixture of 400 mg/l of grit and 2 kg/m³ of shredded cotton and polyester rags (dry weight) with varied widths and lengths up to 1,000 mm x 75 mm (ref: Test 3).  The Main Pump can handle significantly longer and larger rags until repeated recirculation within the test loop causes these to grow to a size capable of blocking the pump suction free passage (ref: test 2, 4 and 5).  The dynamic conglomeration of ropes and clumps that form and ultimately break down (given enough time) cannot be ignored in a high concentration, re- circulating rag test of this nature.  The strategy of starting with smaller rags that will develop into conglomerations of the required size over time is an effective strategy for addressing this dynamic. A ratio in the range of 1:10 (starting to final size) proved effective in the present case.  No clogging or accumulation of rags was seen in the pump casing, at the impeller outlet, or behind the impeller at any time during test program.

Figure 9: KSB GIW SVP-84 Main Pump

All images and graphics used in this article are courtesy of KSB.