Sensitivity Analysis Of Simulation Model Research Articles

Kriging: Methods and Applications

In this chapter we present Kriging also known as a Gaussian process (GP) model which is a mathematical interpolation method. To select the input combinations to be simulated, we use Latin hypercube sampling (LHS); we allow uniform and non-uniform distributions of the simulation inputs. Besides deterministic simulation we discuss random simulation, which requires adjusting the design and analysis. We discuss sensitivity analysis of simulation models, using functional analysis of variance (FANOVA) also known as Sobol sensitivity indexes. Finally, we discuss optimization of the simulated system, including robust optimization.

Fundamentos para el análisis de riesgo climático en zonas áridas

Los sistemas productivos de zonas áridas están sujetos a la aleatoriedad climática lo que les da la categoría de sistemas bajo riesgo. El uso de modelos de simulación que consideren esta aleatoriedad, son la mejor opción para cuantificar la probabilidad de ocurrencia de eventos de lluvia y/o escurrimiento (riesgo). En el presente trabajo se analizan los fundamentos teóricos estadísticos–matemáticos necesarios para incursionar en la simulación y modelación de procesos hidrológicos en zonas áridas. También, se propone un método para cuantificar la precipitación y escurrimiento en sistemas de captación de agua de lluvia para aquellas regiones desprovistas o con poca instrumentación. De igual manera se pondera la necesidad de realizar el análisis de sensibilidad del modelo así como el proceso de calibración.

Sensitivity analysis of simulation model on railway rolling noise prediction

The railway rolling noise prediction software (RRNPS) is a simulation model for predicting the sound pressure levels during a train passage due to wheel/rail roughness, based on vibration dynamics, contact mechanics and sound radiation modules. Some field validation has been done under European and Australian conditions. Comparisons between simulations and measurements have shown that this software model gives reliable predictions in terms of overall A-weighted sound pressure level and noise spectrum. In this paper, based on a typical railway rolling noise situation in Australia (a series of field measurements were taken at a narrow track gauge testing site), a sensitivity analysis of the RRNPS was carried out to investigate the effect of train speed, normal load, sleeper spacing, rail pad stiffness, track gauge and rail cross receptance factor on railway rolling noise. The results provide some insight into model-based methods to control and mitigate railway noise at its source.

A Simple Approximate Semi-analytical Solution for Estimating Leakage of Carbon Dioxide Through Faults

Abstract Assuring CO2 storage security is essential for the widespread implementation of carbon capture and sequestration. Appraising the potential for leakage through faults in seals is an important component of site screening, assessment, and selection. The focus of this study is to understand and quantify the potential rates of CO2 leakage via faults and fractures which could provide fluid migration pathways from the storage reservoir to overlying aquifers. Several analytical solutions exist for estimating rates of fluid migration between reservoirs via faults or leaky wells [1] , [2] , [3] , [4] , [5] , [6] , [7] , [8] . However, there is little focus on leakage up finite length faults. Here we present a new semi-analytical approximate solution for CO2 leakage through a finite length fault zone that relies on a derivation similar to that of calculating the single phase flow rate through a series of units. Under many conditions, this solution provides a good first order estimation of the amount of CO2 that leaks into the overlying aquifer relative to the amount of CO2 injected into the system with only basic knowledge of system geometry and permeability values. Detailed sensitivity analysis of simulation models was performed in order to understand which fault and reservoir parameters most strongly influence leakage rates of CO2 from storage reservoirs. Based on this analysis the three most important parameters were, in order of sensitivity, reservoir permeability, fault permeability and aquifer permeability. With these results, a semi- analytical approximation was developed which relies almost entirely on these permeabilities and the geometry of the system (ie. reservoir and aquifer height, fault thickness, etc.) While this solution does not incorporate multiphase fluid flow properties, it still provides a good approximation for CO2 leakage from a saline aquifer especially when the relative permeability characteristic curves result in mobility ratios near one for typical CO2 saturation values in the plume, which is common for Brooks-Corey relative permeability curves and viscosity ratios for supercritical CO2 and brine at reservoir conditions. Results from this semi- analytical solution are compared to over 50 different numerical models with different fault geometries and locations and a wide range of permeability values for the reservoir, fault and overlying aquifer. Overall, leakage predictions from the analytical solution compare very well with the numerical simulations. The approximation improves when faults are assumed to have no capillary pressure however many cases with capillary pressure are examined. Finally, the approximation is more accurate at lower leakage rates (leakage

Sensitivity Analysis of Simulation Models

This contribution presents an overview of sensitivity analysis of simulation models, including the estimation of gradients. It covers classic designs and their corresponding (meta)models; namely, resolution-III designs including fractional-factorial two-level designs for first-order polynomial metamodels, resolution-IV and resolution-V designs for metamodels augmented with two-factor interactions, and designs for second-degree polynomial meta-models including central composite designs. It also reviews factor screening for simulation models with very many factors, focusing on the so-called 'sequential bifurcation' method. Furthermore, it reviews Kriging metamodels and their designs. It mentions that sensitivity analysis may also aim at the optimization of the simulated system, allowing multiple random simulation outputs.

Sensitivity analysis and stability patterns of two-species pest models using artificial neural networks

Sensitivity analysis is a critical step in mathematical modelling of ecological processes and it provides an idea of the response of the model dynamics to a variation in the values of some parameters. In analytic models, there are standard mathematical techniques for carrying out sensitivity analyses, but this is not so with simulation models, mainly due to the fact that their behaviour usually depends upon the interaction among different parameters, and so sensitivity analysis has to be carried out for all combinations of all parameters of interest. In this study, we explored the use of artificial neural networks (ANN) for sensitivity analysis of simulation models, as applied to simulations models of two-species pest populations: the parasitoid–host system Nezara viridula–Trichopoda giacomellii, N. viridula being a pest of soybean and the Sirex noctilio– Pinus radiata system, S. noctilio being a pest of pine plantations. We compare the ANN sensitivity analysis results with the ones of the Classification Trees (CT), Sobol and the stepwise multiple regression with standardized partial regression coefficients (SMR). The sensitivity analyses were carried out evaluating the simulations models’ parameters effect on the stability behaviour of the simulation models. The ANN sensitivity analysis produced the same (or superior) results as the other two techniques (CT, Sobol and SMR), but showed additional advantages similar to those offered by sensitivity analyses of analytic models: partial derivatives were calculated to determine the contribution of each parameter of the simulation models to their stability behaviour. We conclude that ANN is adequate for simulation modelling sensitivity analysis with the additional advantage of evaluating the contribution of model parameters to the model's behaviour. Although, we used only two-species pest systems as an example, this approach may be applied in wide areas of pest management and population dynamics studies.

Non-linear sensitivity analysis as a tool for explaining population dynamics in a life-system conceptual framework

Life-system analysis is the examination of interacting ecological processes affecting population dynamics. The major problem in application of life-system analysis is in developing methods for analysis of ecological processes. I use a non-linear sensitivity analysis of simulation models to evaluate the role of individual ecological processes and their interactions in population dynamics. The method is applied to the problem of population stability. Coefficients of m- and v-stability characterize stability of the mean and variance of population density, respectively. Contributions of ecological processes and their interactions to m- and v-stability are represented by non-linear sensitivity coefficients. The second application is the analysis of common pine sawfly (Diprion pini L.) population dynamics in Russia. Process interactions are shown to be important in outbreak patterns of this pest.

Techniques for sensitivity analysis of simulation models: A case study of the CO2 greenhouse effect

Sensitivity analysis is needed for valida tion, what-if analysis, and optimization of complicated simulation models. One set of techniques for sensitivity analysis consists of least squares curve fitting, regression analysis, and statistical designs such as factorial designs. In this case study these techniques are applied to several modules of a large integrated assessment model for the greenhouse effect, developed in The Nether lands. The regression models turn out to be valid approximations to the simulation models. Some estimated effects are quite surprising for the simulation users.