Generalized least‐squares parameter estimation from multiequation implicit models

Marshall, Stephanie Pace

doi:10.1002/aic.690491012

Cited by 10 publications

(6 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For batch equilibrium, nonconvergence of gradient method is due to local minima, flat zone of the error function or infinite loop phenomena 5. Extrapolation of these conclusions to inverse modelling of reactive transport lets expect many convergence problems for gradient like methods as assumed by Marshall 6. Even if the Monte‐Carlo methods are more time‐consuming, they are indeed more robust than the gradient‐like methods because they do not use the slope of the objective function.…”

Section: Introductionmentioning

confidence: 99%

Poly(L‐lactide). IX. Hydrolysis in acid media

Tsuji

Nakahara

2002

J of Applied Polymer Sci

115

View full text Add to dashboard Cite

Amorphous and crystallized poly(l-lactide) (PLLA) films are prepared by quenching and annealing at 140°C for 600 min, respectively, from the melt. Their hydrolysis is investigated at pH 2.0 in HCl and dl-lactic acid (DLLA) solutions (37°C) for up to 300 days, and the results are compared with those obtained for PLLA films hydrolyzed at pH 7.4 (phosphate-buffered solution) and pH 12 (NaOH solution). The changes in the weight loss and molecular weight distribution of the PLLA films during hydrolysis in the acid media reveals that their hydrolysis proceeds homogeneously along the film cross section by mainly a bulk erosion mechanism. Moreover, the durability of PLLA films in the acid media is very similar to that in the neutral medium but higher than that in the alkaline medium. The hydrolysis rate constant values (k) of the initially amorphous PLLA0 film evaluated from the changes in the numberaverage molecular weight were 3.0 and 2.4 ϫ 10 Ϫ3 day Ϫ1 at pH 2.0 in the HCl and DLLA solutions, respectively. These k values are very similar to 2.6 ϫ 10 Ϫ3 day Ϫ1 at pH 7.4 in the phosphate-buffered solution. The similar k values and the negligible weight loss after the hydrolysis for 300 days reflect that the hydronium ions and the lactic acid oligomers and monomers have insignificant catalytic effects on the hydrolysis of the PLLA films. Increasing the initial crystallinity of the PLLA film increases the hydrolysis rate in the HCl solution, whereas increasing the initial crystallinity of the PLLA film does not alter the hydrolysis rate in the DLLA solution. The differential scanning calorimetry results show that the crystallization of PLLA chains occurs during the hydrolysis, irrespective of the hydrolysis acid media and the initial crystallinity.

show abstract

Section: Introductionmentioning

confidence: 99%

Poly(L‐lactide). IX. Hydrolysis in acid media

Tsuji

Nakahara

2002

J of Applied Polymer Sci

115

View full text Add to dashboard Cite

show abstract

“…The Gauss-Newton generalized nonlinear least-square algorithm, first described by Britt and Luecke (1973), was employed for parameter estimation. The Gauss-Newton algorithm, which uses exact derivatives, and has been proven to converge in relatively few iterations and provide more accurate estimates of the asymptotic standard errors and covariance, is widely applied to a variety of engineering/scientific studies (Cooley, 1992;Demetracopoulos, 1994;Elmoursi and Gfrerer, 1994;Li et al, 2004;Marshall, 2003). The L 0 value of 322 m 3 /ton for LFG and 167 m 3 /ton for methane was determined, in comparison with Pelt et al (1998) andEnvironment Canada (2001) who have indicated L 0 of 340 m 3 /ton for LFG and 170 m 3 /ton for methane.…”

Section: Landfill Gas Generation Ratesmentioning

confidence: 43%

Performance results from the Tucumán solid waste bioreactor

McBean¹,

Syed-Ritchie

Rovers

2007

Waste Management

View full text Add to dashboard Cite

“…Monte Carlo (MC) simulation is a wide-ranging method to compute statistical characteristics of an output function utilizing a large number of iterations for its random variables. MC simulation has been successfully used for parameter estimation by many researchers, including Zhang and Guay (2002);Marshall, (2003); Aggarwal and Carrayrou, (2006).…”

Section: Monte Carlo Simulation Sampling and Genetic Algorithmmentioning

confidence: 99%