ABSTRACT. The work presented in this article pertains to the use of multibody dynamics to analyze the damage of bearings and gears in the drivetrains of large floating offshore wind turbines. The assessment is performed as a case study on a 10MW offshore reference wind turbine. A notable contribution to the cost of offshore wind turbine maintenance costs stems from the drivetrain. To enhance reliability, this study explores the impact of drivetrain modelling accuracy on the damage estimation of bearings and gears.The investigation is performed using a number of models, some developed in MATLAB through Lagrange’s Equations, and some developed using the commercial multibody dynamics software ADAMS. These models differ in their consideration of body flexibility, gear modelling as well as bearing fidelity, which allows for analysis of the impact of a number of modelling options on drivetrain simulation results. The models span from rigid body dynamics and linear bearings, to considering nonlinear bearings, gears with flexible teeth,and most bodies as flexible. Models are simulated for two load cases, and the results are analyzed. The flexibility of the bedframe was found to have a notable effect on the main and first planetary stage bearings. Bearing clearance was foundto be essential, hence either nonlinear or linear bearings with clearance are essential for accurate damage estimates. In planetary gear stages, where significant non-torque loads may appear, the flexibility of gears and transmission shafts proved to yield notable changes in damage estimates on both gears and surrounding bearings. Despite their simpler formulation, Lagrangian models have a significant computational cost. Nevertheless, they are still competitive as compared with models developed with commercial software. Therefore, these simplified models could be used in preliminary designs since they allow for quick design alterations, before moving on to more detailed analysis.

ABSTRACT. We present a software framework to model and simulate the perception capabilities of rovers operating in extreme lunar environments, enhancing the test-driven development of rover perception systems. Using the multiphysics engine Project Chrono, this framework integrates high-fidelity terramechanics and sensor simulations to model the mechanical and optical properties of navigation and sensing in lunar environments. Additionally, we present POLAR3D, a pipeline for creating photorealistic synthetic images of lunar environments based on NASA's POLAR stereo dataset.

ABSTRACT. When simulation contact interaction in mechanical systems, the contact models require accurate geometric information determined through collision detection methods. Dynamic formulation and collision detection can be more challenging when the mechanical system involves flexible bodies, as the geometric boundaries of such components keep changing during the simulation. The floating frame of reference(FFR) formulation is suitable for flexible systems with small deformation. This work provides a systematic dynamic formulation and time-stepping method for flexible systems with contact using the FFR formulation. In addition, a curve-based collision detection method that is more consistent with the dynamic formulation has been proposed, which can achieve accurate and efficient performance.

ABSTRACT. Usually, detailed impact simulation models within flexible multibody systems have to be set up manually rather than being generated automatically. This is because the process requires prior knowledge of the time and location of the impact, as well as the element resolution within the contact area. If the penalty method is used to determine the occurring contact forces, the corresponding penalty factor also needs to be determined manually. This work, however, presents an adaptive algorithm to simulate impacts within flexible multibody systems fully automatically using reduced isogeometric analysis (IGA) models, the floating frame of reference formulation, and quasistatic contact models for an efficient but still accurate simulation. The adaptive algorithm detects impacts in the system, determines the contact locations on the bodies, refines the contact area, and determines the penalty factor, and therefore automatically simulates impacts. The work shows how to automatically simulate impacts in flexible multibody systems without user action or prior knowledge of impact location and size. As an application example, the setup of two 3D flexible double pendulums is simulated. Since the double pendulum setup is chaotic, the locations of the impacts vary. The goal is to validate the adaptive algorithm and simulate the system for a longer time period with multiple impacts. This is achieved by comparing the results with the analytical solution by Hertz and monitoring the energy of the system.

ABSTRACT. Flexible endoscopes are medical devices that can be modelled as beams due to their slender geometry and non-linear behaviour. When simulating endoscopic procedures, the contact between the beam model and the surrounding has to be considered in the mathematical model. We present a variational formulation of Euler's elastica described by an augmented Lagrangian with unilateral contact potentials. We show preliminary results for the elastica in contact with walls, whose geometry is of particular interest for our industrial application, and give an outlook on possible future implementations.

ABSTRACT. There are different ways to account for contact interactions in a mechanical system; each approach has its own advantages and disadvantages. In this work, we aim to compare two different contact modelling possibilities in a co-simulation setup. Co-simulation involves the simultaneous simulation of multiple interconnected physical systems using different simulation tools. The process of a co-simulation involves defining the interface between different subsystems and exchanging interface variables between them at specific communication points called macro time steps. To determine the variables at the interface between the communication points, we consider two model-based modellings which involve creating a reduced interface model (RIM) of the mechanical system to mimic the behaviour of the full model at the interface. The first method, the smooth RIM, assumes that the contact state will remain unchanged during the macro time step. The second method, the non-smooth RIM, involves identifying potential contact pairs and accounting for them through the solution of a linear complementarity problem during the time step, allowing for the possibility of contact attachments or detachments between communication points. Using a robotic arm as an example, it was found that the smooth RIM can produce inaccurate results in certain cases. One solution to this issue is to regularize some of the constraints by adding constitutive relations, but that can change the physics the model represents, and determining the proper stiffness and damping coefficients can be challenging. In contrast, the non-smooth RIM was found to produce results that were more accurate and in line with the reference solution without changing the system model.

ABSTRACT. Simulation and simulation-assisted methods have a large impact for product development in many industrial fields. Multibody dynamics (MBD) simulation, for instance, is an established method used for durability analysis and energy efficiency calculations in vehicle engineering, based on its ability to provide accurate prediction of interaction forces. In the domain of off-road vehicles and heavy machinery, such considerations need to be expanded by a method to model the soil and its interaction with the vehicle, i.e., the soil-tool interaction, as the resulting forces may drastically influence the durability and energy efficiency of the machine. Therefore, the model must be chosen carefully to maintain the high accuracy in the force prediction for the soil-tool interaction, as required for a reliable and robust product development.  

In this contribution, we present a workflow to tackle this topic, based on a co-simulation scheme between an MBD-based vehicle model and a particle simulation realized in the Discrete Element Method (DEM). The simulated soil is parametrized and validated by matching simulation results from a virtual experiment with measurement data from real-world soil laboratory experiments as the triaxial compression test. Using this process, the applicability and performance of the numerical methods can be determined.

ABSTRACT. This work presents a novel framework for the experimental model identification of flexible multibody mechanisms. It is shown that by exploiting the flexible natural coordinate formulation (FNCF) and vision-based measurements, it becomes possible to use a least-squares model identification method without the need for time-intensive model simulations between optimizer iterations. By using the FNCF formulation, both the mass matrix and the stiffness matrix in the equations of motion are constant. This property opens up the possibility for a least-squares model identification where, instead of the often time-consuming model simulations between optimizer iterations, simple matrix multiplications are used. However, this requires knowledge of the full vector of generalized coordinates and the external excitation for each time step. By only using conventional sensors (e.g., accelerometers, strain gauges, encoders), this is difficult to achieve, especially for flexible multibody mechanisms where measurements of both the rigid body motion and the flexible deformations are required. Therefore, this research proposes the use of vision-based measurements, as they can provide full-field motion measurements of the mechanism with sufficient accuracy and spatial density to extract individual component deformations. The vision-based motion tracking in this research uses an affine Lucas-Kanade optical flow in combination with Procrustes motion separation to obtain the components' rigid body motions and deformation motions. Both a hybrid modal decomposition and a singular value decomposition are exploited to decompose the deformation motion into individual modes and participation factors. As a validation platform, a planar slider-crank mechanism is used. Here, the crank is considered a rigid component while the connecting rod is assumed to be flexible. The setup is recorded with a Photron SA-Z High-Speed Camera at 40,000 frames per second while the crank rotates at a speed of 60 rad/s. The external excitation is obtained by a torque readout from the driving servomotor.

ABSTRACT. This article delves into the validation of continuous friction models for use in sensitivity analysis and optimization of dynamic systems with friction. Simulation of friction and stiction effects in dynamic systems is predominantly done through two types of friction models: quasi-static and dynamic models. Quasi-static models are continuous approximations of the discrete Coulomb phenomenon, consisting of the stiction phase and the dynamic friction phase. On the contrary, dynamic friction models are event-based differential equations aimed at identifying the instantaneous friction phase and producing the appropriate friction force. Dynamic models tend to be more accurate representations of real-world friction but are much more complex to implement, especially in sensitivity analysis and optimization case studies. This article presents preliminary investigations into the validity of using quasi-static models for such scenarios. The Rabinowicz experiment is used as a benchmark model for parameter and signal estimation case studies using the quasi-static Brown McPhee model and the dynamic event-based Coulomb model. Initial results show that quasi-static models are well-suited for sensitivity analysis and optimization case studies.

ABSTRACT. The development of a multibody vehicle model, generated using the MotionSim and integrated into to the Simulink environment, has been described in this study. MotionSim is a collection of codes, which has been developed in the Matlab, specifically to model multibody systems without physical definitions utilizing general coordinates, as well as to allow one the ability to modify the source code to include any special constraints or elements that may be required.

MotionSim code has then been integrated into to the Simulink environment, providing a pictorial representation of physical signals and methods. With the flexibility of Matlab’s toolboxes and MotionSim being native codes, integrating Artificial Intelligence into multibody simulations can easily be achieved. In this study, the actual tire data by the Calspan Institute, providing tire testing data to Formula SAE teams, has been fitted to neural network models, which predict the tire lateral and longitudinal forces based on the slip angle and ratio.  All this information is put together and can be simulated in the Simulink environment.

The general multibody solver is encapsulated into the full vehicle Simulink model block. This allows the integration of control systems into the model, which can include simple elements such as a driver model utilizing a simple PID loop that may control the vehicle in the specified scenario.  It also allows the inclusion of control and simulation of active suspension elements such as continuously variable dampers and stability control.

ABSTRACT. With the increasing demand for lighter vehicles, to reduce the CO2 emissions and extend the autonomy of electric cars, it is particularly relevant to understand the effect of lighter and softer materials on the dynamic behaviour of road vehicles. The use of alternative structural joints to bolting and spot-welding techniques is fundamental to join dissimilar materials and composites. Suitable numerical methods are essential to better understand how materials and structural joints employed in the construction of the body-in-white (BiW), which is the main structural component of a road vehicle, affect ride, handling and active safety. However, it is still not clear how realistic changes in the materials and structural joints of the BiW of a car, to reduce weight and increase life cycle, may affect the ride, handling, and active safety. The present work discusses the development of flexible multibody models of road vehicles, considering the inherent tire-road contact phenomena, complex suspension systems, and component flexibility, to examine the effects of changing the materials and structural joints of the BiW in the dynamic behaviour of road vehicles.

ABSTRACT. When modelling geometrically exact beams under constraints, two problems arise:The presence of nonlinear configuration spaces for describing large rotations and the presence of algebraic variables coupled with differential ones. To obtain an efficient numerical solution for the latter problem, half-explicit Runge-Kutta methods have been introduced in 1992 by Brasey and Hairer. The current work adapts these half-explicit Runge-Kutta methods to solve DAEs in nonlinear configuration spaces, of the Lie group form, so that the first problem is covered. The study aims at the analysis of the so called drift-off effect, which is a deviation from the constraints at position level when evaluating the numerical solution of the index-2 DAEs. Specifically, we adapt classical techniques for avoiding the drift-off effect to nonlinear configuration spaces. To conclude the study, numerical experiments on flexible structures modelled as Cosserat rods are performed.

ABSTRACT. This contribution describes the development of a computational model for the rope-sheave contact interaction in reeving systems when the ropes are modeled with an arbitrary Lagrangian-Eulerian approach. This discretization approach has been developed in previous publications as a general and systematic method for the modeling and simulation of reeving systems. However, the rope-sheavecontact model was avoided assuming the no-slip contact condition. The contact model developed in this paper introduces specialized ALE-ANCF-cubic rope contact elements that are used to discretizethe rope segment winded at the sheave. The contact is modeled using a set of virtual discrete bristles attached to material points in the mid-line of the rope in one end and in contact with the sheave in the other end. Therefore, a second Lagrangian mesh, apart of the ALE mesh used todiscretize the rope, is used to define the fixed ends of the bristles. The kinematics and dynamics used to calculate the normal and tangential contact forces are described in detail. The contact model is 3D and can be used to analyse the contact with a sheave groove with arbitrary shape. The tangential contact force model can be used to describe stick and slip contact conditions and, to improve the simulation performance of the model, a LuGre regularization tangential contact force model is used. The rope-sheave contact model is used to analyze the behavior of a simple elevator system. The numerical results show that the static rope-sheave contact interaction agrees well with an analytical solution of the problem. Finally, the same elevator system is analyzed dynamically for a cabin ride of 8 meters with a steady velocity of 1 m/s. Results show that the normal and tangential contact forces during the steady velocity period are not so different to the static solution, but very different of the classical Creep Theory and Firbank’s Theory.

ABSTRACT. To geometrically exact describe the deformation of a Cosserat beam, it is modeled as a curve in SE(3). From a kinematic point of view, the crucial task is to infer this curve merely from the boundary conditions. In this paper a cubic and quartic interpolation scheme is presented. The interpolation takes initial and terminal values of the body-fixed strain measure as input. These 3rd/4th-order interpolation scheme allows exactly reconstructing the displacement of a beam (with constant cross section or cross linearly changing cross sections) subjected to a general wrench applied at the beam.

ABSTRACT. A general method for modelling flexible multibody systems based on finite elements is applied to the analysis of deflections of leaf springs as used in precision machinery. First, the method is reviewed. Then, a finite beam element for modelling leaf springs that can include effects of constrained warping, non-linear torsion and anticlastic curvature is presented. The formulation makes use of assumed strains that are integrated over the length of the beam. A comparison of large static deflections of a cantilever leaf spring is made with a previous formulation and detailed finite element solutions. An application to a parallel leaf spring guidance flexure is shown.

ABSTRACT. Autonomous bipedal robot locomotion is a challenging task because it requires stable, robust, and highly dynamic walking. Although our humanoid robot LOLA is capable of stable and robust walking in uncertain environments, its motion is far from being highly dynamic. A precise analysis of its walking motion is required to better understand the underlying gait pattern and to enhance the robot's soft- and hardware capabilities to achieve a highly dynamic gait. In this talk, we present an approach to compute and analyze the underlying fundamental gait pattern of LOLA using Proper Orthogonal Decomposition (POD). Human gait serves as an ideal model for walking robots, and thus a comparison with the basic human gait pattern is performed based on our POD-based gait characterization.

ABSTRACT. The autonomous cooperation of multiple mobile robots allows for solving larger and more complex tasks than a single robot can handle. Since robotics is about automating mechanical work, investigating mechanically challenging model problems is crucial for developing control concepts in distributed robotics. So far, transporting rigid objects has been a popular benchmark problem, where often a fixed number of robots or specific object types are assumed. Based on our previous work, which copes with the transportation task utilizing a versatile distributed optimization-based scheme explicitly controlling the agents' postures and exerted forces, it is the goal to investigate whether multiple simple mobile robots are also able to cooperatively manipulate highly flexible objects.Therefore, a multibody simulation is set up in which the robots shall cooperatively manipulate a highly flexible sheet that is rigidly attached to them. To consider nonlinear effects, which appear to be crucial to manipulate the sheet, e.g., to elevate it, the flexible body is modeled using ANCF. Moreover, the considered problem is also interesting from a control point of view since the agents cooperatively have to control different goals like exerting forces to obtain a desired object shape while moving their formation along a predefined path.

ABSTRACT. Networks of systems are of increasing relevance, in particular in robotics, where a cooperating and communicating network of robots can achieve things unachievable by a single robot. However, while networks of systems with underlying distributed decision-making algorithms bring increased flexibility, they can also be vulnerable. In particular, in uncontrolled environments, it may happen that malevolent agents enter the network, derogating the performance of the whole network. To deal with such scenarios, this contribution proposes two methods to identify anomalous behavior in robotic networks, one of which is based on a model-driven, inverse optimal-control approach, whereas the other uses machine learning in a stochastic framework. To analyze the proposed methods, two well-known problems from distributed robotics, namely formation control and the coverage problem, are considered. The contribution shows that both methods can yield very high detection rates of anomalous behavior, which ranges from merely erroneous to actively antagonistic behavior.

ABSTRACT. This study proposes simulation-based deep learning models for fast predicting the impact accelerations of the spacecraft during soft landing on the complex airbag landing system. The finite element model was constructed to generate the dataset with multiple inputs and outputs. The deep learning models were developed under the framework of the multi-layer perceptron (MLP) and the convolutional neural network (CNN), respectively. We first trained separate MLP models for different impact accelerations and then organized one CNN model with time-stepping to output all accelerations simultaneously.  By comparing predictions on longer time series, the CNN model demonstrates better extrapolation than the MLP model. The results indicate that deep learning models can improve the efficiency of system design optimization and real-time prediction. 

ABSTRACT. We describe a framework that can integrate prior physical information, e.g., the presence of kinematicconstraints, to support data-driven simulation. Unlike other approaches, e.g., Fully-connected NeuralNetwork (FCNN) or Recurrent Neural Network (RNN)-based methods that are used to model the sys-tem states directly, the proposed approach embraces a Neural Ordinary Differential Equation (NODE)paradigm that models the derivatives of system states. A central part of the proposed methodology is itscapacity to learn the multibody system dynamics from prior physical knowledge and constraints com-bined with data inputs. This learning process is facilitated by a constrained optimization approach, whichensures that physical laws and system constraints are accounted for in the simulation process. The mod-els, data, and code for this work are available at https://github.com/jqwang2373/PNODE-for-MBD.

ABSTRACT. This research uses machine learning techniques to enhance a feedforward controller for a fully actuated 2 degrees of freedom manipulator with flexure joints. The foundation of the controller is a combination of the Lagrangian Neural Network to model the system’s conservative forces and the Feedforward Neural Network to simulate non-conservative ones. To address the limitations of both networks in precisely modeling the reproducible part of these forces, we introduce the weighted least-squares method with regularization, which maps the system’s configurations to the residue of control signals (error) and adjusts the model with rank-1 updates. Inevitable trade-offs apply when one uses Time-Delay Embedding, but the preliminary results indicate its feasibility in application to improve the used error learning approach.

ABSTRACT. Large Language Models (LLMs) are currently experiencing a high level of attention due to their extraordinary skills. They are experiencing strong growth and new applications with high benefits for industry and business are added every day. In this work, we investigate the current capabilities of LLMs related to the generation of multibody dynamics models from natural language inputs. In particular, we investigate LLMs that have been trained on our specialized multibody code, Exudyn, and which are able to capture and translate the complexities of kinematics and dynamics into functional programming interfaces. Additionally, we investigate the fine tuning of open-source LLMs from HuggingFace and compare the performance with closed-source models.

ABSTRACT. Co-simulation environments are simulation setups in which two or more dynamic solvers are coupled by exchanging of a limited set of variables at specific communication points in time. When co-simulation is performed according to an explicit (noniterative) coupling approach, numerical errors are introduced and the quality of the results is compromised. We provide a data-driven framework to infer subsystem dynamics in explicit co-simulation environments to make the co-simulation much less vulnerable to errors in total energy.

ABSTRACT. For activities with high-intensity demands where muscle fatigue and the loss of muscle force are anticipated, the integration of muscle fatigue models with muscle force and load-sharing paradigms becomes increasingly important. In a recent work, the authors integrated multi-level models to account for redundant muscle forces within a multibody environment, along with the three-compartment controller (3CC) muscle fatigue model. While their results provided reasonable estimates, they observed that the 3CC model did not accurately capture the expected force decay during the training session, thus suggesting that the activity could be sustained indefinitely. This observation appears to be unrealistic and contradicts findings from other muscle fatigue studies. In this study, to enhance the simulation of intermittent short-duration high-intensity exercises, the authors opted for modifying the 3CC muscle fatigue model. Specifically, they introduced a four-compartment model that distinguishes between the short-term fatigued state (corresponding to metabolic inhibition) and the long-term fatigued state (simulating central fatigue and potential microtraumas).

ABSTRACT. The authors aimed to simulate the contact between the patella and the femur after total knee replacement. Due to the high-performance computing requirements, rigid-body multibody dynamics formulations were applied. A comparative analysis between two collision detection methods employed to simulate interactions between rigid bodies was conducted. The first method is a mesh-to-mesh collision detection algorithm, which discretizes the bodies into triangular mesh surface elements. The second method employs analytical surface expressions to offer closed-form solutions for addressing the contact problem.

ABSTRACT. The muscles wrap around their underlying bone segments in human musculoskeletal system, and the obtained curved muscle paths strongly influence their nonlinear dynamic responses. Researchers often simplify muscle wrapping as the shortest path problem without considering time-dependent muscle-bone contact. In comparison, conventional finite element method (FEM) of skeletal muscles introduces expensive computational costs. To describe nonlinear muscle-bone wrapping with acceptable computational cost, a flexible multibody modeling of the skeletal muscle was established based on absolute nodal coordinate formulation (ANCF). It was further utilized in evaluating the human-machine interactions of wearing a space suit based on forward dynamics simulations.

ABSTRACT. Popular formulations for multibody systems rely on unit-quaternions (Euler-parameters) for tracking orientations of rigid bodies. Consequently, numerical integration on quaternions also carries the normalization constraint on quaternions. Recent developments in multibody dynamics have demonstrated efficient solution methods for differential algebraic equations of frictionless constrained motion in absolute coordinates.  The state-of-the-art methods propose Lie-integration directly on the orientation matrix, thus circumventing the need to carry normalization constraint on quaternions. However, extracting accurate and unique orientation histories from the rotation matrix either requires numerically solving a set of nonlinear equations, or finding eigenvalues of the rotation matrix. Both of these processes incur additional computing cost. The presented study proposes a rotation-preserving exponential integration scheme that operates directly on quaternions. Further, we propose modifications in state-of-the-art differential variational inequality (DVI) framework for modeling friction in smooth systems using absolute coordinates. While DVI formalism is usually adopted for systems with contacts, simulating smooth systems with ideal joints with friction makes DVI method a universal formalism for rigid-body systems. The proposed method poses the general problem as an optimization problem with equality constraints, and emphasizes on resolution of the constraint forces into contact forces, to then evaluate the friction forces. The derived formulations are demonstrated through numerical simulations on three fundamental mechanical joints. Further, the simulation results of the proposed integration scheme are compared with those of state-of-the-art Lie-integration on rotation matrices. The successful implementation of the described approach further highlights the versatility of the DVI approach.

ABSTRACT. Martian thin atmosphere renders atmospheric flight significantly more challenging than flight on Earth. On the contrary to the conventional rotary and fixed wing configurations that have limited utilization due to their inefficiency in Low Reynolds flight regime, atmospheric flight on Mars by using flapping wing propulsion/lift gaining technique is a promising concept. To this end, the objective of the paper is to present a numerically efficient computational model and optimisation-based validation tool that will allow for ‘automated’ testing of the flapping-wing design principles how to fly on Mars. The algorithm is based on reduced-order FSI (fluid-structure interaction) model on Lie groups and quasi-steady model for a fruit fly-like aerial vehicles, combined with the Discrete Mechanics and Optimal Control (DMOC) approach.

ABSTRACT. This work presents the development of a Multibody model of an aircraft seat assembly based on the plastic hinge technique approach to improve the seat structural performance and assist in the Certification by Analysis.

The structural components within the seat’s primary load path - the legs, the spreaders, and the cross tubes, were segmented into several bodies. To simulate the elastic and plastic deformations of the seat, the bodies were connected by plastic hinges. The constitutive models of the hinges were obtained from the Finite Element Analysis that were performed on the bodies.

The MB model was validated by comparing the loads at floor fittings and seat-belt attachments, as well as the kinematics and relevant injury criteria to the occupant (ATD) obtained in the MB model, to the ones from experimental procedures for the dynamic certification tests.