IMPACT
For analyzing the driving performance of pile foundations
Overview
IMPACT is used for analyzing the driving performance of pile foundations using a 1-dimensional wave equation model. The interaction between pile and soil is captured by non-linear springs and dashpots distributed along the pile shaft and base. The pile-soil springs are based on theoretical continuum models and, because the internal soil column is treated explicitly with separate interaction springs compared with the external soil, it is particularly suitable for large diameter open ended ‘monopiles’ used to support offshore wind turbines. The application also includes the unique ability to be run in a ‘data driven mode’, whereby the soil resistance to driving (SRD) is computed directly from input cone penetration test (CPT) data. Users have the ability to select from a range of widely used SRD models for particular soil types or specify custom options.
Soil Resistance to Driving
Data from cone penetration tests (CPT) are ideal for assessing the ‘static resistance to driving’ (SRD) of a pile as well as for evaluating long-term axial capacity after installation and equalization. Pile-soil interaction parameters can be derived from the cone data, including values of end-bearing resistance and maximum values of shaft friction relevant in the current vicinity of the pile tip.
To streamline the workflow in performing pile driving analysis, IMPACT includes the ability to import a digital CPT record. IMPACT automatically assigns a soil type using one of the available classification methods (e.g., Robertson 1990, Schneider et al. 2008, Robertson 2016).
Components of the SRD model are then defined with respect to each of the soil types. The SRD interface is broken into end bearing and shaft resistance components. To account for friction fatigue, the shaft resistance component is further broken into parts that define the maximum shaft resistance (tmax) in the vicinity of the pile tip, a ratio of the minimum degraded shaft resistance to the maximum (tmin/tmax) and a degradation rule according to the current distance of the soil element from the pile tip. Together, these components define the actual shaft resistance (ts) at a given point along the pile shaft for a specific pile embedment. Each of these components are defined under a menu item in the image below. Given there is no general parametric equation that can be used to define these components for all current (and potential future) SRD models, a novel ‘ string to equation’ interface was developed, which enables users to type equations that define each component of the SRD model.
The table below lists the variables that can be used in the equations to define the SRD model. Examples in the figure above show equation strings used to represent the Alm and Hamre SRD model. Because friction fatigue (or degradation) is included in the SRD model definition, multiple pile embedment depths can be analyzed in a single run and the blowcount for each penetration depth can be calculated very rapidly.
Most SRD models do not include an internal shaft resistance component due to the soil plug. However, recent studies on large diameter open ended piles have indicated that improved back analysis of driving records can be achieved by including an additional component of resistance due to the soil plug. Therefore, IMPACT provides an option to specify the ratio of the internal soil plug resistance (tp) to the external shaft resistance (tp/ts).
For situations were CPT data is not available, or where finer control over the SRD profile are needed, IMPACT also allows for the "manual" specification of the SRD model by pasting spread sheet data into an excel like table in the interface.
Hammer Models
IMPACT includes tabulated information for a wide range of hammers, with details provided of ram weight, nominal energy etc. For a given analysis, these may be modelled explicitly, by using the first ‘segment’ of the pile as the ram, with other segments then representing cushion and anvil and finally the pile. The blow is initiated by specifying an initial ram velocity v0, which is automatically assigned to all nodes of the ram segment. Connections between the various segments representing the hammer are specified as compression-only, therefore allowing hammer bounce.
More conveniently, however, an analytical approximation to the downward force generated by a given hammer may be used (Deeks and Randolph 1993). This requires specification of the ram and anvil masses, as well as the stiffness and damping of the cushion between ram and anvil.
Stress Wave Matching
The Stress Wave Impact FiT (SWIFT) function was developed to assess the quality of signal matching of stress waves at a given depth. It is broadly similar in concept to the CAPWAP match score, although is based on a scale of 0 to 100 with 100 being a perfect match and has a different weighting when assessing the match quality.
The SWIFT weighting is a function of normalized time t/(2L/c), which avoids limiting the match to 25 ms after the return time. This may prove problematic, since the match score depends slightly on pile length and type. However, for offshore piles the stress wave perturbations can continue for a long time and the match weighting needs to extend far enough to capture the soil response near the base of the pile.
Parallel Computing
IMPACT is a scalable web based applications that runs on Amazon Web Services. The web interface is hosted on a regular virtual web server. However, the core IMPACT calculation algorithm is run on AWS as a Lambda function. AWS Lambda is a serverless compute platform, which creates and runs an instance of the IMPACT function in response to each calculation request. There is no practical limit to the number of parallel simulations Lambda can run. Within the IMPACT web interface, users can submit up to 150 calculation sequences. For example, because IMPACT has a built in SRD algorithm, multiple pile depths can be analysed in a single run. For each depth, the web server creates separate input files that trigger independent Lambda functions. Therefore all depths are analysed in parallel. This results in significant speed up of the calculation time. For example, a single IMPACT simulation may take 20 seconds. If 100 depths were run in series, this would take over 30 minutes to compute. With the IMPACT cloud architecture, each Lambda function takes around 0.5 seconds to trigger in series. Therefore, the simulation for the 100th depth would start around 50 seconds after the first depth and all calculations would be complete 20 seconds after this. Therefore, more than 30 minutes of calculations can be completed in just over 1 minute.
To further exploit this architecture, an application programming interface (API) has been made available. This allows users to ‘script’ simulations with IMPACT while achieving the same computational speed up generated on the web interface.
