Michael G. Parsons

FORECASTING TANKER FREIGHT RATE USING NEURAL NETWORKS

Improvement in forecasting accuracy is a difficult task but critical for business success. This paper investigates the potential of neural networks for short- to long-term prediction of monthly tanker freight rates. Procedures are outlined for the development of the neural networks. The problem of under-training and over-training is addressed by controlling the number of iterations during the training process of neural networks. A comparative study of predictive performance between neural networks and ARMA time series models is conducted. Our evience shows that neural networks can significantly outperform time series models, especially for longer-term forecasting.

An assessment of fuzzy logic vessel path control

This paper proposes a novel approach to modeling the four quadrant dynamic response of thrusters as used for the motion control of ROV and AUV underwater vehicles. The significance is that these vehicles are small in size and respond quickly to commands. Precision in motion control will require further understanding of thruster performance than is currently available. The model includes a four quadrant mapping of the propeller blades lift and drag forces and is coupled with motor and fluid system dynamics. A series of experiments is described for both long and short period triangular, as well as square wave inputs. The model is compared favorably with experimental data for a variety of differing conditions and predicts that force overshoots are observed under conditions of rapid command changes. Use of the model will improve the control of dynamic thrust on these vehicles.

A finite element analysis of beams on elastic foundation including shear and axial effects

Abstract A displacement finite element method for analyzing a beam on continuous elastic foundation is presented. A three-dimensional model which accounts for the effects of both the Filonenko-Borodich and Pasternak foundation models in a consistent and complete way is used. A variational principle is introduced with the slope field due to bending only and the displacement field approximated by independent quantities subjected to variation. Numerical examples illustrate the accuracy of the element, the importance of shear, axial and shear-axial interaction effects associated with continuous elastic foundation, and finally the application of the element to a rotor supported by two hydrodynamic journal bearings.

Probabilistic Torsional Vibration Analysis of a Marine Diesel Engine Shafting System: The Level Crossing Problem

In this analysis, the shafting shear stress is found to be a Gaussian, harmonizable cyclostationary process with a harmonic series representation consisting of two complex conjugate components. In this paper, the level crossing problem for this stress process is studied

Integrated electric plants in future Great Lakes self-unloaders

The feasibility and potential benefits of using Integrated Electric Plants in future Great Lakes self-unloaders are evaluated. Integrated Electric Plants, the all-electric ships, utilize electrical propulsion motors and central station power generation that powers all propulsion, thruster, self-unloading equipment, and other ship service needs. Inte- grated Electric Plants have become the plant of choice in many recent naval vessels, cruise ships, high technology cargo vessels, and special purpose vessels, such as offshore supply and service vessels and icebreakers. This study considers arrangements, effects on cargo capacity, fuel usage, and environmental emissions in all operating modes, maintenance requirements, and manning. The comparison is made for two notional self-unloading bulk carriers: a 1000 ft Poe Lock maximum self-unloader and a 730 ft MacArthur Lock, Welland Canal, St. Lawrence Seaway maximum self-unloader.

Optimal Commonality Decisions in Multiple Ship Classes

A methodology is presented for the determination of the Pareto optimal choice of components and elements to make common between two different classes of military vessels. The use of commonality can produce fleet-wide savings in component purchasing, training, spare parts, vessel construction, etc. The methodology presented here determines the optimal commonality decision and designs the vessel classes to maximize the mission performance per average acquisition cost of each vessel class and the total fleet saving achieved by the commonality. A customized evolutionary algorithm is used to determine the resulting discrete Pareto surface. The methodology is illustrated by its application to the design of two ship classes to perform the specific missions of the US Coast Guard’s National Security Cutter and Offshore Patrol Cutter. The results show that the methodology is effective and that not all commonality choices produce a net savings.

Applications of Optimization in Early Stage Ship Design

Recent research at the University of Michigan developing and applying modern optimization methods to early ship design decision making is reviewed. These examples illustrate the use of fuzzy logic, genetic and evolutionary algorithms, and agent methods to solve complex multicriterion ship design problems. The first application optimizes an early stage hull form for both smooth water powering and seakeeping performance using an advanced evolutionary algorithm taking into consideration the change of vessel weight with the hull form variation. The second application supports the optimization of naval ship general arrangements using a new hybrid agent-genetic algorithm method and stochastic generation algorithm. The final example uses an evolutionary algorithm to establish the optimal commonality to use in two ship classes that are to share components and features in order to save overall fleet costs. These show how these advanced ship design methods can be used to aid early ship design decisions.

An initial assessment of fuzzy logic vessel path control

A fuzzy logic controller for ship path control in restricted waters is developed and evaluated. The controller uses inputs of heading, yaw rate, and lateral offset from the nominal track. A Kalman filter is used to produce the input state variables from noisy measurements. The controller produces a commanded rudder angle. Input variable fuzzification, fuzzy associative memory rules, and output set defuzzification are described. The controller is benchmarked against a conventional linear quadratic Gaussian (LQG) optimal controller and Kalman filter control systems. An initial startup transient and regulator control performance with an external hydrodynamic disturbance are evaluated. The fuzzy controller yields competitive performance.

The potential conversion of the U.S. Great Lakes steam bulk carriers to liquefied natural gas propulsion : initial report

The feasibility and potential benefits of converting 10 remaining U.S. flag Great Lakes steamship bulk carriers to liquefied natural gas (LNG) propulsion using gas engines is investigated. This is the final report of a study that was initially reported at the February 24, 2012, SNAME Section Meeting in Cleveland, Ohio. The evolving marine air emissions standards and the movement to LNG fuel in U.S. and international non-LNG carriers and the general case for the possible conversion of the remaining U.S. flag Great Lakes steamship bulk carriers to LNG fuel were outlined in the initial report. In this sequel, the final results of a conceptual design study on the conversion of the three AAA class vessels (SS Arthur M. Anderson, SS Cason J. Callaway, SS Philip R. Clarke), focusing primarily on operational and arrangement feasibility and remaining life-cycle economics, are presented. Three cases, a pure diesel conversion, a single-fuel LNG conversion, and a dual-fuel LNG/diesel conversion, are compared.

A Hybrid Agent — Genetic Algorithm Approach to General Arrangements

Abstract The continuing development of an agent-genetic algorithm hybrid optimization kernel for the allocation of spaces to Zone-decks in a preliminary ship general arrangements design system is presented. In this system, design agents representing the interests of the spaces and the Zone-decks arc utilized to propose possible improvements to a population of candidate solutions. Their design change requests arc accepted or denied by a domain agent if they will improve the overall ship space allocation design. In parallel, a design judge agent monitors the population and reseeds dominated or infeasible solutions and a genetic algorithm agent provides periodic stochastic manipulation of the population to improve the algorithm’s global search capabilities. The algorithm is used to optimize the Zone-deck allocation of spaces for a Notional Corvette-sized vessel.