Design of experiments for discrimination of rival models based on the expected number of eliminated models

Alberton, André L.; Schwaab, Marcio; Lobão, Marcos Wandir Nery; Pinto, José Carlos

doi:10.1016/j.ces.2012.03.010

Cited by 8 publications

(3 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Typically, some available data may be exploited at this stage to choose among competitive modeling approaches through suitable discrimination techniques. , At least in the easiest cases a χ 2 test on experimental data may be sufficient to make the discrimination. ,, Ad hoc experiments can also specifically be designed to allow for a more effective and reliable discrimination among different candidates. − Once a suitable candidate model has been selected (and a preliminary estimation of its parameter has been carried out), the model may need upgrading to improve its capability of representing the phenomena being investigated (new experiments may be needed and possibly designed, and an estimation of all model parameters should be attempted).…”

Section: Model Developing Approachmentioning

confidence: 99%

An Identifiable State Model To Describe Light Intensity Influence on Microalgae Growth

Bernardi

Perin

Sforza

et al. 2014

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

Despite the high potential as feedstock for the production of fuels and chemicals, the industrial cultivation of microalgae still exhibits many issues. Yield in microalgae cultivation systems is limited by the solar energy that can be harvested. The availability of reliable models representing key phenomena affecting algae growth may help designing and optimizing effective production systems at an industrial level. In this work the complex influence of different light regimes on seawater alga Nannochloropsis salina growth is represented by first principles models. Experimental data such as in vivo fluorescence measurements are employed to develop the model. The proposed model allows description of all growth curves and fluorescence data in a reliable way. The model structure is assessed and modified in order to guarantee the model identifiability and the estimation of its parametric set in a robust and reliable way.

show abstract

Section: Model Developing Approachmentioning

confidence: 99%

An Identifiable State Model To Describe Light Intensity Influence on Microalgae Growth

Bernardi

Perin

Sforza

et al. 2014

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

show abstract

“…Significant efforts have been invested in the development of different discrimination 84 criteria for experimental designs in different fields [42][43][44][45][46][47][48][49][50][51][52][53][54], but only a few criteria have been 85 applied to groundwater modeling, e.g., the differences between model outputs [38,39,41,55] and 86 the Kulback-Leibler information [42]. 87…”

Section: Introduction 38mentioning

confidence: 99%

Bayesian experimental design for identification of model propositions and conceptual model uncertainty reduction

Pham

Tsai

2015

Advances in Water Resources

View full text Add to dashboard Cite

12The lack of hydrogeological data and knowledge often results in different propositions (or 13 alternatives) to represent uncertain model components and creates many candidate groundwater 14 models using the same data. Uncertainty of groundwater head prediction may become 15 unnecessarily high. This study introduces an experimental design to identify propositions in each 16 uncertain model component and decrease the prediction uncertainty by reducing conceptual 17 model uncertainty. A discrimination criterion is developed based on posterior model probability 18 that directly uses data to evaluate model importance. Bayesian model averaging (BMA) is used 19 to predict future observation data. The experimental design aims to find the optimal number and 20 location of future observations and the number of sampling rounds such that the desired 21 discrimination criterion is met. Hierarchical Bayesian model averaging (HBMA) is adopted to 22 assess if highly probable propositions can be identified and the conceptual model uncertainty can 23 be reduced by the experimental design. The experimental design is implemented to a 24 groundwater study in the Baton Rouge area, Louisiana. We design a new groundwater head 25 observation network based on existing USGS observation wells. The sources of uncertainty that 26 create multiple groundwater models are geological architecture, boundary condition, and fault 27 permeability architecture. All possible design solutions are enumerated using a multi-core 28 supercomputer. Several design solutions are found to achieve an 80%-identifiable groundwater 29 model in five years by using six or more existing USGS wells. The HBMA result shows that 30 each highly probable proposition can be identified for each uncertain model component once the 31 discrimination criterion is achieved. The variances of groundwater head predictions are 32 significantly decreased by reducing posterior model probabilities of unimportant propositions. 33 34 Keywords: Bayesian model averaging, experimental design, groundwater, model discrimination, 35 observation network design, uncertainty. 36 37The lack of hydrogeological data and knowledge often results in different propositions 60 (or alternatives) to represent uncertain model components and creates many candidate 61 groundwater models using the same data. For example, geological architecture can be one 62 uncertain model component in groundwater modeling. Many hydrostratigraphy modeling 63 techniques may be proposed to construct different geological architectures (propositions) 64 potentially leading to an overwhelming number of models with non-dominant posterior model 65 probabilities. Conducting prediction and uncertainty analysis using a great deal of 66 computationally intensive groundwater models can become intractable. By incorporating many 67 conceptual models, prediction results using BMA can become useless when the prediction 68 uncertainty is very high. This concern highlights the importance of conducting an experimental 69 design to discriminate g...

show abstract

“…A better approach would probably make use of optimal sequential experimental design techniques based on the kinetic models derived for esterification or transesterification reactions. As an initial approach, a sequential design for model discrimination might be performed, using one of the many approaches found in the literature, ,− followed by a sequential design for precise parameter estimation, such as the D-optimal designs, if an improvement in the parameter estimates is needed. , …”

Section: Resultsmentioning

confidence: 99%

Are Empirical Models Based on the Response Surface Methodology Suitable for Biodiesel Production Optimization?

Kober

Schwaab

Barbosa-Coutinho

et al. 2022

Ind. Eng. Chem. Res.

Self Cite

View full text Add to dashboard Cite

The response surface methodology (RSM) can constitute a useful tool for the optimization of biodiesel production when applied to establish the relationship between biodiesel yields and the remaining process operation variables. For this reason, in the present study, several kinetic models were fitted to three sets of experimental data available in the literature and compared to second-order empirical models traditionally used in RSM studies. The obtained results show the asymptotic increase of oil or free fatty acid (FFA) conversion (X A) with the increase of the reaction time, alcohol to oil/FFA molar ratio, and catalyst concentration, differently from the behavior predicted by most empirical models described in the literature and used to perform RSM studies. Particularly, when the reaction temperature is also varied, it is possible to observe experimental conditions where X A attains a maximum value as the result of the nonlinear structure of the kinetic models proposed in the present work. Therefore, these results clearly show that phenomenological models should be preferred during the optimization of biodiesel production and also used to perform RSM studies.

show abstract

Design of experiments for discrimination of rival models based on the expected number of eliminated models

Cited by 8 publications

References 22 publications

An Identifiable State Model To Describe Light Intensity Influence on Microalgae Growth

An Identifiable State Model To Describe Light Intensity Influence on Microalgae Growth

Bayesian experimental design for identification of model propositions and conceptual model uncertainty reduction

Are Empirical Models Based on the Response Surface Methodology Suitable for Biodiesel Production Optimization?

Contact Info

Product

Resources

About