Assessment of Wind Turbine Foundation Response Using Field Instrumentation and Dynamic Laboratory Testing of Unsaturated Site Soil
Abstract
EXECUTIVE SUMMARY
Wind turbine generator (WTG) foundations undergo continuous, dynamic loading conditions that make their design unique and complex. Though many foundation alternatives exist for WTG’s, the most common design is a shallow, octagonal, gravity-based footing. Two utility-scale (≥ 1.5 MW) WTG’s in the Midwest were outfitted with foundation and tower instrumentation to monitor in situ loads. Furthermore, undisturbed soil samples were taken at the site and tested with resonant column and cyclic triaxial devices that were retrofitted with suction-saturation control capabilities. This thesis focuses on the analysis and interpretation of field instrumentation data and laboratory results to assess current WTG foundation design practices.
For field instrumentation, nine pressure gauges (PG’s) and ten soil deformation gauges (SG’s) were placed under the foundation to measure the distribution of contact pressure and vertical soil deformation. At Site A, four thermal dissipation sensors (TDS’s) were installed in the soil below the footing to measure fluctuations in water content. In addition, six sets of full-bridge strain gauges were placed on the WTG tower at two elevations to measure the tower loading (i.e. wind induced moment and tower vibration). Finally, three MEMS accelerometers were installed on the foundation pedestal to measure foundation tilt.
Regarding laboratory testing, a Hardin-type resonant column and load-controlled cyclic triaxial apparatus were outfitted with suction-saturation control capabilities. New bottom platens for each were machined with a hole for a high air entry (HAE) ceramic disc to control the flow of water in and out of the specimens. Using the axis translation technique, the soil suction could be controlled by increasing the air pressure relative to the pore water pressure. Four conditions were tested: 0 kPa (saturated), 25 kPa, 50 kPa, and 100 kPa soil suction. To supplement the data, soil water characteristic curves (SWCC’s) were developed to relate the soil suction and soil saturation.
Soil suction (i.e. saturation) is known to impact the stiffness of the soil. Stiffness is a crucial soil property in the design of a WTG foundation, as it controls the response of the structure to the dynamic loads imparted by the oscillating tower. Previous analysis of these WTG sites has indicated that the foundation is out-performing its expected stiffness response. One hypothesis is that the soil, being unsaturated, is stiffer than engineers might assume based on assumptions for saturated soil. By understanding the soil’s response to varying saturation, implications to the foundation design and performance can be drawn.
Tower moment and shear force were found to be correlated with wind speed, with magnitudes within expected design values. Furthermore, the moment along the predominant wind direction (PWD) was found to spike when the wind direction coincided with the PWD. When overturning moment was compared to the soil deformation and vertical pressure, the three mirrored each other extremely well, indicating a predictable, consistent load transfer. The overturning moment and pressure along the PWD also proved to have a nearly linear relationship. When normalized to the static, zero wind-load vertical pressure, the pressure responses at the two sites were nearly identical. The windward and leeward pressure gauges also nearly perfectly mirrored each other, indicating that the windward uplift was roughly balanced by the leeward pressure spike. Finally, the harmonic analysis of the tower and soil pressure response revealed that the amplitude of the cyclic moment was about an order of magnitude less than the overall overturning moment, and that the cyclic pressure amplitude was nearly two orders of magnitude less than the static pressure.
The application of matric suction to the two approaches proved to be effective overall in affecting the stiffness response. A clear trend of increasing shear modulus with increasing soil suction was present. Moreover, this increase existed at all strain magnitudes measured, though the effect was most pronounced at low strain. At a soil suction of 100 kPa, the value of G0 more than doubled in many specimens. The method worked better with the resonant column, as it is non-destructive, so the suction could be progressively increased without changing the specimen out. Thus, the results of the resonant column are likely more representative of the true response of a soil to changes in saturation, whereas the cyclic triaxial method requires a new specimen for each suction level, adding in the variability of soil properties along the profile. Furthermore, the variation in response across the site was significant, with G0,sat varying from 43 to 76 MPa. Based on the boring logs, the site subsurface was reported as very uniform. The use of undisturbed soil, however, showed that, though the site appeared uniform, the laboratory tested soil stiffness distribution did not necessarily follow the same trend.
Using the data from one of the thermal dissipation sensors, TDS-1, the in situ water content (Θ) directly below the footing could be assessed. The sensors revealed a clear seasonal trend, where moisture content reaches a peak of 0.4531 in November. The minimum recorded value occurred in minimum of 0.4515 in August. Since the void ratio where the sensors were placed is unknown, assumptions from measurements at other site locations were made. Assuming the maximum measured Θ corresponds to the saturated condition, the stiffness within the range of measured saturation levels ranged from approximately 43 MPa to 115 MPa for specimens tested at a confining stress of 70 kPa, approximately the confining stress at TDS-1.
Overall, the results of this study can apply in a few key ways. First is through understanding of the impacts of soil suction on the stiffness. A shallow foundation situated 10 m above the water table, for example, would theoretically feel roughly 100 kPa of suction at its base, which could more than double the stiffness. Furthermore, this increase in soil stiffness could be compounded by a two-orders-of-magnitude decrease in shear strain, which would double the reduced shear modulus for this soil type. Altogether, these observations could result in a reduction in the required radius for rotational stiffness to the point where it no longer controls the overall design.