Portland Oregon's Willamette River West Side CSO (Figure 1), described in more detail in a preceding article,¹ included a 970 000 m³ per day (220-mgd) pump station at Swan Island.  The Swan Island Pump Station Shaft (or the "shaft") design required a 40-m (130-foot)-diameter excavation to a depth of about 49 m (160 feet).  At the shaft location, geologic conditions were predominantly sand and gravel soil materials (fill, alluvium, and Troutdale formations) of relatively high permeability. 

A cut-off wall system was installed around the shaft perimeter to minimize groundwater inflow during excavation.  The cut-off wall included:      

• A 61-m (200-foot)-deep cylindrical slurry wall
• A jet grout wall extending downward another 38 m (125 feet) and keyed into a low permeability mudstone formation (Sandy River Mudstone). 

Dewatering was required to lower groundwater levels within the shaft to allow excavation of soil and construction of the pump station.  The required drawdown inside the shaft was 43 m (140 feet).

PB Monitored Shaft Performance Testing

The contractor was tasked with implementing a series of performance tests to assess hydraulic behavior of the wall during dewatering.  Excess leakage of groundwater into the shaft during dewatering could lead to multiple constructability and safety issues, including shaft wall and base stability.  The performance testing plan involved installation of dewatering wells within the shaft interior along with multiple water level monitoring devices (both piezometers and multi-level transducer arrays) inside and outside of the shaft.  On behalf of the City, PB reviewed and monitored the contractor's shaft performance testing.

Initial Performance Testing

Phase 1 Test.  After initial construction of the cut-off wall system, the contractor's team carried out a Phase 1 performance test at a low dewatering rate.  The results indicated that a significant amount of leakage was entering the shaft. A pumping rate of 1 m³ (250 gallons) per minute resulted in:         

• Approximately 5.5 m (18 feet) of water level drawdown inside the shaft
     
• An observed drawdown of about 1.5 m (5 feet) in the soils surrounding the shaft wall. 
     

Under those cut-off wall performance conditions, it was estimated that a peak pumping rate of more than 9 m³ (2,000 gallons) per minute would be necessary to achieve the 43 m (140 feet) of drawdown within the shaft required for construction. 

Phase 2 Test.  A more detailed Phase 2 performance test was run at a higher pumping rate of 7.4 m³ (1,680 gallons) per minute in an attempt to identify the leakiest areas of the cut-off wall.  The multi-level transducers installed inside and outside of the wall allowed for some measurement of the horizontal and vertical variability of groundwater head throughout the soil formation. 

At the 7.4-m³ (1,680-gallon) per minute pumping rate, a maximum interior drawdown of about 27 m (90 feet) was observed near the bottom of the slurry wall where a majority of the pump/well intake occurred.  Outside the shaft the maximum measured drawdowns, which varied from 8 m to 12 m (27 feet to 40 feet), were observed at the sensor located about 3 m (10 feet) below the junction of the slurry wall and the jet grout wall. 

Significant drawdowns were observed at all the deep sensors adjacent to the jet grout wall, while only very small drawdowns were observed at the shallow sensors outside the slurry wall.  While an unacceptable volume of leakage was occurring, according to these results the slurry wall appeared to be performing well, and most of the leakage into the shaft was likely occurring through the jet grout.  Also, based on the radial variability of observed drawdown surrounding the shaft, it appeared that more leakage may have been occurring on the northern side of the shaft, in the vicinity of wall Panel 17.

Groundwater Flow Modeling

Using the Phase 2 test data, a numerical groundwater flow model was created to assess wall leakage conditions and investigate several hypotheses regarding the distribution of flow into the shaft.  The finite-difference MODFLOW code developed by the USGS (McDonald and Harbaugh, 1988) was used to build the model.       

• Horizontal and vertical conductivity values in the model were adjusted during calibration to obtain a reasonable fit to head values observed during the Phase 2 testing. 
• Lower values of hydraulic conductivity were assigned to the cylindrical ring of model cells surrounding the shaft to represent the slurry and jet grout cut-off wall system. 

The flow model was used to perform various simulations to test the effects of different inflow scenarios (Figure 2).  The results led to the following conclusions:

• If the slurry/jet grout wall system was fully effective, the anticipated inflow across the underlying Sandy River Mudstone would be on the order of 1.3 m³ (300 gallons) per minute.      
• The slurry wall appeared to be much more effective that the jet grout columns; however, this did not mean that there couldn't be isolated weak spots in the slurry wall.
• The jet grout appeared to allow leakage at various locations around its entire perimeter but appeared to be leakier in its top half, closer to the junction with the slurry wall.
• A very small percentage of the total inflow appeared to be coming across the mudstone.

Remediation and Additional Performance Testing

Following the first round of grout remediation efforts developed and implemented by the contractor, a Phase 3 performance test was implemented at a pumping rate of approximately 6.9 m³ (1,570 gallons) per minute.  Results showed improved performance, with more interior drawdown and less exterior drawdown achieved with a slightly lower dewatering rate.  The maximum observed drawdown inside the shaft was about 35 m (115 feet), which was still less, however, than the required 43 m (140 feet).

The contractor team planned additional remedial grouting of the wall and installation of a greater number of water level monitoring instruments prior to the start of excavation.  Following the remediation, the Phase 4 pump test was run.  A pumping rate of 6.4 m³ (1,450 gallons) per minute achieved maximum interior drawdowns of up to 40 m (130 feet), which closely approached the required goal. 

A comparison of the water level data from the Phase 2, 3, and 4 testing (Figures 3 and 4) showed stepwise improvement in cut-off wall performance following the two successive jet grout remediation efforts.  Drawdown inside the shaft increased with successive phases (Figure 3), while drawdown outside the shaft decreased in successive phases (Figure 4), even though the pumping rate for each consecutive test was less than the previous rate.

Conclusion

Results of the Phase 4 performance test were positive and indicated that a workable condition had been achieved.  Actual excavation and construction of the Swan Island Pump Station Shaft began in February 2004 (Figure 5).  The use of detailed water level monitoring, analysis, and modeling benefited the overall project by facilitating assessment of cut-off wall performance and allowing for rational recommendations regarding areas of focus for remediation, leading to a final constructed project for the City of Portland.