심해저 채광용 무한궤도차량의 주행 신뢰성 설계

Abstract

The objective of this study is to design a tracked vehicle of a deep-sea manganese nodule mining system for securing the reliability of vehicle mobility crawled on extremely soft and cohesive soil, for instance, seabed in Korea reserved 5 area of Clarion-Clipperton fracture zone. The design characteristics of the tracked vehicle on deep-seabed are investigated since the cohesive and weak terrain at seafloor significantly affects the mobility of the tracked vehicle. The large variation, unknown probability distribution, and a few sample data of shear strength for seafloor sediment are the distinguishing features. Thus, the reliability-based design for mobility would be applied in order to design systematically the tracked vehicle. The probability distribution of shear strength is estimated from the sample data and that of steering ratio is assumed as an exponential distribution in design formulation. The 3 design variables such as interval of tracks, slenderness ratio and weight of buoyancy module, and 2 noise random variables such as shear strength and steering ratio are chosen. The mobility is classified into flotation and traction which are represented by sinkage and slip, respectively. The performance functions are defined as the discrepancy of response functions and each capacity. The target reliability for each performance function is selected as 0.999. The weight of the vehicle is chosen as an objective function. From the sample data of multiple corer by Pacific ocean explorations of Onnuri research vessel, the shear strength has been measured by a shear vane at 2 cm depth interval from seafloor surface between 1997 ~ 2006. The vehicle contact surface (VCS) is defined as the surface which the tracked vehicle would touch potentially after the uppermost soil is removed by water-jet lifting device assembled in front of the tracked vehicle. The probability distribution of the shear strength is parametrically and nonparametrically estimated for 2 ~ 22 cm depth from VCS. The combined distribution of generalized Pareto distribution in both tails and piecewise linear estimation in body, one of the nonparametric estimation methods, is chosen as the best probability distribution with respect to the test statistic of the Kolmogorov - Smirnov test. The sinkage and slip are calculated by using dynamic simulation software of the tracked vehicle crawled on soft soil. Reece model for pressure-sinkage relation, the correlation model for traction-shear displacement relation, and the modified Bode model for slip sinkage-shear displacement are suggested and applied as track-soil interaction model. Kriging metamodel is constructed for 3 design variables and 2 noise random variables for calculating efficiently the responses. The reliability is calculated by using the Monte Carlo simulation. Descriptive sampling is used for random number generation. The optimum design is obtained for the defined design problem. The interval of tracks is decreased from initial value. The steering ratio is increased to upper boundary. The weight of buoyancy module is increased from initial value. The reliabilities for sinkage and slip are approached to the required target reliabilities. The objective function, the weight of the vehicle, is decreased by 6.6 % from the weight of initial design. Hence, the results for steering crawling are compared with those for straight crawling. In straight crawling, we learn that the slenderness ratio of contact area significantly affects the slip. Piecewise linear estimation (PLE) among various statistical analyses methods shows better performance, and PLE is simpler to be applied in this study as well. In conclusion, optimum design of the lighter tracked vehicle with satisfying the target reliabilities of mobility is obtained. This study in R&D stage will be effectively applied to design systematically a commercial mining tracked vehicle. |본 연구는 태평양 심해저 망간단괴 채광용 무한궤도차량의 주행 신뢰성 설계를 수행하는 것을 목적으로 한다. 태평양 광구의 심해저면은 연약하고 점착성이 강한 지반으로, 그 위를 주행하는 무한궤도차량의 성능은 해저지반 강도에 많은 영향을 받는다. 또한, 해저면 강도로 대표되는 환경인자뿐만 아니라 선회비와 같은 작동인자도 주행성능에 큰 영향을 준다. 이에, 심해저면 작업 중에 무한궤도차량의 유지보수와 모니터링이 어려운 특징으로 인해, 확률론적 설계접근을 통해 설계단계에서 체계적으로 신뢰성을 확보할 필요가 있다. 좌우 무한궤도차량의 간격, 무한궤도 접지면 길이에 대한 폭의 비(세장비), 부력재 모듈 무게 등을 설계변수로 선정하였다. 해저면 전단강도와 좌우 무한궤도차량의 속도비(선회비)를 잡음확률변수로 선정하였다. 무한궤도차량의 주행성능은 부양성능과 견인성능으로 분류될 수 있다. 부양성능의 지표인 침하와 견인성능의 지표인 슬립을 응답함수로 선정하였다. 한편, 차량 무게를 목적함수로 선정하였다. 심해저면 전단강도의 확률분포를 모수적 및 비모수적으로 추정하였다. 장기간에 걸친 대양 조사선 탐사에서 얻은 전단강도 데이터를 확보하였다. 본 데이터를 바탕으로 단괴 채집시 물제트에 의해 해저면 일부가 제거된 이후 차량이 직접적으로 접촉할 면을 차량접촉면이라고 정의하고, 이러한 차량접촉면 아래 깊이 별로 전단강도 데이터를 재정리 하였다. 비모수적 추정방법들 중 하나인 generalized Pareto distribution과 piecewise linear estimation(PLE)을 결합한 추정방법이 Kolmogorov-Smirnov검증의 통계치를 기준으로 할 때 전단강도 분포특성을 가장 잘 표현하는 확률분포임을 알 수 있었다. 침하와 슬립은 TRACSIM(Tracked vehicle simulation)이라는 연약지반무한궤도차량 동역학해석 프로그램을 이용하였다. 이때 침하특성, 견인특성, 슬립침하특성 등 트랙과 지반간 특성은 각각 Reece 모델, 상관관계 모델, 수정된 Bode 모델 등 회귀모델을 사용하였고 분할트랙시험을 통해 각 모델의 파라메터를 산출하였다. 크리깅 근사모델을 통해 주어진 설계변수와 잡음확률변수에 대한 침하와 슬립 근사모델을 구축하였다. 한편, 본 연구에서는 침하와 슬립의 신뢰도를 추정하기 위해 몬테카를로 시뮬레이션을 적용하였다. 최적설계를 수행하여 초기치 대비 6.6 % 경량화되고 침하와 슬립의 성능함수 신뢰도가 목표 신뢰도를 만족하는 최적치를 구하였다. 직진주행과의 비교를 통해 세장비가 직진에서는 슬립을 감소시킴을 알 수 있었다. 신뢰도 해석 성능과 적용편의성을 고려했을 때 여러 통계분석 방법중 piecewise linear estimation이 가장 합리적인 분석방안으로 제안된다. 본 연구를 통해 추후 상업용 집광 무한궤도차량 개발 시에 체계적으로 채광 시스템을 설계할 수 있는 토대가 마련된 것으로 판단된다.; The objective of this study is to design a tracked vehicle of a deep-sea manganese nodule mining system for securing the reliability of vehicle mobility crawled on extremely soft and cohesive soil, for instance, seabed in Korea reserved 5 area of Clarion-Clipperton fracture zone. The design characteristics of the tracked vehicle on deep-seabed are investigated since the cohesive and weak terrain at seafloor significantly affects the mobility of the tracked vehicle. The large variation, unknown probability distribution, and a few sample data of shear strength for seafloor sediment are the distinguishing features. Thus, the reliability-based design for mobility would be applied in order to design systematically the tracked vehicle. The probability distribution of shear strength is estimated from the sample data and that of steering ratio is assumed as an exponential distribution in design formulation. The 3 design variables such as interval of tracks, slenderness ratio and weight of buoyancy module, and 2 noise random variables such as shear strength and steering ratio are chosen. The mobility is classified into flotation and traction which are represented by sinkage and slip, respectively. The performance functions are defined as the discrepancy of response functions and each capacity. The target reliability for each performance function is selected as 0.999. The weight of the vehicle is chosen as an objective function. From the sample data of multiple corer by Pacific ocean explorations of Onnuri research vessel, the shear strength has been measured by a shear vane at 2 cm depth interval from seafloor surface between 1997 ~ 2006. The vehicle contact surface (VCS) is defined as the surface which the tracked vehicle would touch potentially after the uppermost soil is removed by water-jet lifting device assembled in front of the tracked vehicle. The probability distribution of the shear strength is parametrically and nonparametrically estimated for 2 ~ 22 cm depth from VCS. The combined distribution of generalized Pareto distribution in both tails and piecewise linear estimation in body, one of the nonparametric estimation methods, is chosen as the best probability distribution with respect to the test statistic of the Kolmogorov - Smirnov test. The sinkage and slip are calculated by using dynamic simulation software of the tracked vehicle crawled on soft soil. Reece model for pressure-sinkage relation, the correlation model for traction-shear displacement relation, and the modified Bode model for slip sinkage-shear displacement are suggested and applied as track-soil interaction model. Kriging metamodel is constructed for 3 design variables and 2 noise random variables for calculating efficiently the responses. The reliability is calculated by using the Monte Carlo simulation. Descriptive sampling is used for random number generation. The optimum design is obtained for the defined design problem. The interval of tracks is decreased from initial value. The steering ratio is increased to upper boundary. The weight of buoyancy module is increased from initial value. The reliabilities for sinkage and slip are approached to the required target reliabilities. The objective function, the weight of the vehicle, is decreased by 6.6 % from the weight of initial design. Hence, the results for steering crawling are compared with those for straight crawling. In straight crawling, we learn that the slenderness ratio of contact area significantly affects the slip. Piecewise linear estimation (PLE) among various statistical analyses methods shows better performance, and PLE is simpler to be applied in this study as well. In conclusion, optimum design of the lighter tracked vehicle with satisfying the target reliabilities of mobility is obtained. This study in R&D stage will be effectively applied to design systematically a commercial mining tracked vehicle.