A novel feasibility analysis method for black‐box processes using a radial basis function adaptive sampling approach

Feasibility analysis is used to determine the feasible region of a multivariate process. This can be difficult when the process models include black‐box constraints or the simulation is computationally expensive. To address such difficulties, surrogate models can be built as an inexpensive approximation to the original model and help identify the feasible region. An adaptive sampling method is used to efficiently sample new points toward feasible region boundaries and regions where prediction uncertainty is high. In this article, cubic Radial Basis Function (RBF) is used as the surrogate model. An error indicator for cubic RBF is proposed to indicate the prediction uncertainty and is used in adaptive sampling. In all case studies, the proposed RBF‐based method shows better performance than a previously published Kriging‐based method. © 2016 American Institute of Chemical Engineers AIChE J, 63: 532–550, 2017     

A unified algorithm for flowsheet optimization

A flexible algorithm for flowsheet optimization is developed on the FLOWTRAN process simulator. The optimization strategy combines the feasible and infeasible path approaches as well as the simpler black box approaches. While the most efficient optimization strategy is often problem dependent, this paper presents guidelines that show which strategy is more efficient for a given problem. Also embedded within the algorithm is a new Broyden strategy for efficiently converging even complex flowsheets, without computing a new Jacobian. This allows for intermediate strategies between the infeasible and feasible path procedures. A ratio test based on the Kuhn-Tucker convergence test automatically and adaptively adjusts the optimization strategy.The implementation on FLOWTRAN is discussed in detail and a number of examples are run to illustrate the flexibility of the implementation as well as demonstrate the effectiveness of the adaptive optimization strategy.     

Derivative‐free optimization for expensive constrained problems using a novel expected improvement objective function

In this work, an algorithm for the optimization of costly constrained systems is introduced. The proposed method combines advantages of global‐ and local‐search algorithms with new concepts of feasibility space mapping, within a framework that aims to find global solutions with minimum sampling. A global search is initially performed, during which kriging surrogate models of the objective and the feasible region are developed. A novel search criterion for locating feasibility boundaries is introduced, which does not require any assumptions regarding the convexity and nonlinearity of the feasible space. Finally, local search is performed starting from multiple locations identified by clustering of previously obtained samples. The performance of the proposed approach is evaluated through both benchmark examples and a case study from the pharmaceutical industry. A comparison of the method with commercially available software reveals that the proposed method has a competitive performance in terms of sampling requirements and quality of solution. © 2014 American Institute of Chemical Engineers AIChE J, 60: 2462–2474, 2014     

Synthesis,characterization, and permeation properties of polyether-based polyurethanes

A system of synthesis of polyether-based urethanes was developed which had sufficient flexibility in composition so that transport properties could be optimized. Mixtures of poly(oxyethylene) glycol (PEG) and poly(oxypropylene) glycol (PPG) of a variety of molecular weights were tied together by varying amounts of kinds of “hard segments.” Thus, the water swell, the mechanical properties, and the size of the soft blocks and hard blocks could be varied. With a fixed content of hard segments, the water absorption decreased with decrease in the PEG/PPG ratio, demonstrating the feasibility of producing controlled changes in hydrophilicity of the polymer without significant change in the mechanical strength. Some polyurethanes based on PEG 600 and PPG 425 had a very good high value of P_w/P_s but a somewhat low value of P_w. The polyurethanes prepared by using phenylenediamines as chain extenders had markedly enhanced modulus and an extended rubbery plateau region, as anticipated.     

Wet-Oxidative Desulfurization as a Flexible and Greener Way for H2S Removal from Acid Gases

In wet-oxidative gas desulfurization processes, absorbed H₂S is directly converted to sulfur by a redox catalyst dissolved in a circulating scrubbing solution. High flexibility is achieved through the availability of different catalyst classes. In this work, a process based on the Takahax process was investigated in a pilot plant under industrially relevant conditions with coke oven gas. A stable plant operation without sulfur precipitation was achieved by a suitable choice of the operating conditions. The solid-free operating window was determined and a sulfur redissolution strategy developed.     

Optimization strategy for encapsulation efficiency and size of drug loaded silica xerogel/polymer core‐shell composite nanoparticles prepared by gelation‐emulsion method

It has been commonly discovered that reducing particle size always accompanies with undesirable deterioration of drug encapsulation efficiency in double emulsion based techniques. However, a clear optimization strategy for process variables to minimize this negative impact has been rarely reported. To fill this gap, we have successfully developed an optimization strategy for silica xerogel/polymer composite nanoparticles prepared by our recently developed gelation‐emulsion method. To develop this strategy, interactive effects of multiple process variables were investigated through a four‐factor three‐level experimental design by considering all screened dominant process variables influencing particle size and encapsulation efficiency, including sonication time of second emulsion (t₂), sonication power of the second emulsion (P₂), total volume of the second emulsion (V₂) and volume ratio of aqueous phase and primary emulsion (r). The optimization strategy for fabricating the target particle size with optimal encapsulation efficiency was designed by adjusting the process variables in the order of r, V₂, t₂, and P₂ With this strategy, conspicuous enhancement of the encapsulation efficiency (e.g., from 27 to 63% for a particle size of 211 nm) and significant increment of the feasible size range through our novel fabrication method from 192–569 nm to 90–914 nm have been achieved in this study. POLYM. ENG. SCI., 2017. © 2017 Society of Plastics Engineers     

A fluidized bed catalytic cracking regenerator model

A grid model including thermal effects is proposed. The aim is the simulation of a fluidized catalytic cracking regenerator similar to the industrial unit of Destileria La Plata, YPF, Argentina. It is demonstrated that a simple C.S.T.R. model without bypass of gas feed entering the bed provides a good approach for representing the fluidized bed including the grid region. In addition, by means of the C.S.T.R. model, it is shown that there exist two characteristic operating regions: a zone where (C^o_o – C_o) depends on the initial coke concentration and a zone where (C^o_o – C_c) is controlled by oxygen supply.     

Continuous fractionation and solution properties of PIB. I. Search for the best mixed solvent and first results of the continuous polymer fractionation

To adopt a recently developed method for large scale fractionation (CPF = continuous polymer fractionation, a special kind of counter current extraction) to polyisobutylene (PIB), a systematic search for the best mixed solvent was performed. For this purpose, the essential parts of the phase diagrams solvent/nonsolvent/PIB were determined for 21 mixed solvents by cloud-point measurements; with eight systems of special interest, the molecular weight distributions of the polymers contained in the coexisting phases were also studied. On the basis of these experiments and of considerations concerning additional criteria for the performance of the continuous counter current extraction, the mixed solvent toluene/methyl ethyl ketone was chosen. First experiments with a PIB sample of M_w = 420,000 g/mol and a molecular nonuniformity U = (M_w/M_n) ? 1 of 2.3 yielded two high molecular weight fractions M_w = 1.1 × 10⁶ and 0.6 × 10⁶ resp., with U = 0.3 on a 100 g scale upon the application of four CPF steps.     

Continuous computer controlled production of formate dehydrogenase (FDH) and isolation on a pilot scale

A continuous production process has been developed up to pilot scale (300 l) for FDH production with the methylotrophic yeast Candida boidinii. A high cell mass specific FDH activity (50 U/g) is achieved by process computer controlled supply of pure methanol to operate the reactor at an optimum methanol concentration of 10 g/l. The maximum FDH spacetime yield achievable with this process control involves a residence time of 7 h. The FDH space-time yield (STY) and FDH concentration are a function of the oxygen transfer rate (OTR) of the fermenter (maximum STY = 255 U/(l h) at k_La = 870 l/h). For a reasonable compromise between high FDH space-time yield and high FDH concentration, an optimum residence time is adjustable by regulating the supply of nutrient salt solution in relation to the OTR of the fermenter. On a pilot scale (200 l continuously stirred tank reactor) roughly 4 million U of FDH were produced within 10 days at a residence time of 14.3 h. Isolation of intracellular FDH enzyme was performed using extraction with an aqueous two-phase system (PEG/K₂HPO₄). A technical product quality of 1.2 U/mg FDH was achieved without any chromatographic purification step.     

Mixing length equation for high schmidt number mass transfer at solid boundaries

A simple modification of the Van Driest damping factor for mixing length calculations of mass and/or heat transfer coefficients at solid boundaries is proposed to increase mass eddy diffusivities in the near-wall region, without affecting velocity predictions. Calculated mass transfer rates with a constant Sc_t = 0.85 are in good agreement with present and previous experimental results obtained in pipe and duct flows at high Schmidt numbers.     

Polymer optical fiber for near infrared use

Using the empirical simulation that we developed to estimate loss spectra of amorphous polymers, we predicted the intrinsic loss spectrum of fluorine-containing polyarylate (FPAr) that should have a small attenuation loss in the near infrared region. Loss of FPAr is smaller than that of poly (methyl methacrylate), which is a typical optical polymer used as a core material in polymer optical fiber (POF), above 720 nm. FPAr was synthesized by phase transfer catalyzed polycondensation of 2,2-bis (4-hydroxyphenyl)-1,1,1,3,3,3-hexafluoropropane with isophthaloyl dichloride. Its glass-transition temperature (T_g) by DSC is 163°C. This indicates that FPAr has higher heat resistance than polycarbonate (PC) by about 10°C. A film of 100μm thick has over 85% transmittance from 330 to 900 nm. Values from a tension test are similar to those of PC. As a result, FPAr is expected to be a good core polymer for POF, because of its high T_g, good transparency, and good flexibility. Next, the POF made of FPAr was fabricated by a melt-flow spinning method. It had larger attenuation loss than predicted. The difference from the simulation was extrinsic loss caused by the fabricating process. Therefore, when a more suitable fabricating process is developed, loss of FPAr-POF can be decreased, so that it will be applicable to optical communications in the near infrared region. © 1993 John Wiley & Sons, Inc.     

Dielectric relaxation in vinylidene fluoride–hexafluoropropylene copolymers

The molecular mobility in copolymers of vinylidene fluoride–hexafluoropropylene VDF/HFP of 93/7 and 86/14 ratios has been investigated by means of broadband dielectric relaxation spectroscopy (10^?1–10⁷ Hz), differential scanning calorimetry DSC (?100 to 150°C), and of wide angle X‐ray diffraction WAXS. Four relaxation processes and one ferroelectric‐paraelectric phase transition have been detected. The process of the local mobility β‐ (at temperatures below glass transition point) is not affected by chemical composition of the copolymer and the formed structure. Parameters of segmental mobility in the region of glass transition (α_a‐relaxation) depend on the ratio of comonomer with lower kinetic flexibility. α_c‐relaxation is clearly observed only in VDF/HFP 93/7 copolymer, which is characterized by a higher crystallinity and a higher perfection of crystals of α‐ (α_p‐) phase. Diffuse order–disorder relaxor type ferroelectric transition connected with the destruction of the domains in low‐perfect ferroelectric phase in the amorphous regions has been detected for both copolymers. An intensive relaxation process (α‐process) was observed for both copolymers in high‐temperature region. DSC data shows that it falls on the broad temperature region of α‐phase crystals melting. It is considered to be connected with the space charge relaxation. © 2007 Wiley Periodicals, Inc. J Appl Polym Sci 2007     

Modelling of chain twist boundaries in poly(vinylidene fluoride) as a mechanism for ferroelectric polarization

It is assumed that the process of ferroelectric polarization of the beta phase of poly(vinylidene fluoride) (PVF₂) in response to the action of the external electric field in direction perpendicular to the molecular axis and to the film, involves movement of the chain twist boundaries. These boundaries, at which every chain is twisted by 180 degrees, separate domains of opposite polarization. The energy barriers that are surmounted as the boundary was advanced one repeat unit were calculated and compared with the energy gained by reversing the polarization of an unfavourably oriented repeat unit in an electric field that produces polarization in PVF₂. It is suggested that the movement of chain twist boundaries, in contradistinction to previously postulated models in which only one chain is twisted at a time, provides a model for the poling of PVF₂ that is feasible energetically and kinetically. Theoretical modelling, analogous to that for Bloch wall that separates domains in magnetic materials, suggest that the process of polarization might be described either as a diffusion process or as the propagation of a soliton along the chains.     

Influence of drying conditions and plasticizer-to-binder ratio on the water-based tape casting of La0.6Sr0.4Co0.2Fe0.8O3−δ

Tape casting is a flexible technique for manufacturing scalable ceramic sheets. This study fabricated La_0.6Sr_0.4Co_0.2Fe_0.8O_3−δ (LSCF6428) tapes using water-based tape casting. A high molecular weight plasticizer, polyethylene glycol 4000, was chosen to balance flexibility and mechanical strength. Adjusting the plasticizer-to-binder ratio (R-value) and increasing relative humidity during drying led to crack-free and flawless green tapes of 330 µm. The uniform polymer matrix improved homogeneity and consistency as well. An applicable suspension formulation was developed for the water-based fabrication of LSCF tapes for continuous production.     

Quantification of process flexibility via space projection

Flexibility index is an effective indicator for characterizing the operational flexibility of a process design model. In this work, the concept of δ-space projection is proposed and defined from the view of solution space, and the flexibility index can be quantified without solving any optimization problems. First, the flexibility index problem is reformulated as an existential quantifier model. Then, the projection operator in the cylindrical algebraic decomposition (CAD) method is introduced to project the solution space onto the one-dimensional feasible space of the flexibility index. Subsequently, all the candidate values of the flexibility index, which may refer to the locations of the critical points, that is, vertexes and tangency points, are extracted. Last, two δ-checking rules based on the traditional and improved CAD methods are proposed for finding the flexibility index. The case studies show that the proposed method can exactly find the flexibility index, regardless of convex or nonconvex systems.     

Predictive Dynamic Model of Air Separation by Pressure Swing Adsorption

A general dynamic model is developed for separation of air over a carbon molecular sieve and a zeolite adsorbent for production of nitrogen and oxygen. The proposed model is validated using experimental data from working laboratory scale N₂–PSA and laboratory scale O₂–PSA systems. Simulations studies are performed to investigate the effect of changing various process variables, such as the duration of PSA steps, bed length and feed inlet velocity.     

Development of thermally conductive polymer matrix composites by foaming‐assisted networking of micron‐ and submicron‐scale hexagonal boron nitride

Thermally conductive polymer matrix composite (PMC) foams with effective thermal conductivities (k_eff) higher than their solid counterparts have been developed for the first time. Using a material system consists of low density polyethylene and micron‐scale or submicron‐scale hexagon boron nitride platelets as a case example, this article demonstrates that foaming‐assisted filler networking is a feasible processing strategy to enhance PMC's k_eff, especially at a low hBN loading. Parametric studies were conducted to identify the structure‐to‐property relationships between foam morphology (e.g., cell population density, cell size, and foam expansion) and the PMC foam's k_eff. In particular, there exists an optimal cell size to maximize the PMC foam's k_eff for foams with up to 50% volume expansion. However, an optimal cell size is absent for PMC foams with higher volume expansion. X‐ray diffraction (XRD) analyses reveal that both the presence of hBN platelets and foam expansion promoted the crystallization of LDPE phase. Moreover, the XRD spectra also provide evidence for the effect of foam expansion on the orientation of hBN platelets. Overall, the findings provide new directions to design and fabricate thermally conductive PMC foams with low filler contents for heat management applications. © 2015 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2016 , 133, 42910.     

MICROFABRICATION OF CYLINDRICAL ION TRAP MASS SPECTROMETER ARRAYS FOR HANDHELD CHEMICAL ANALYZERS

A new microfabrication method for micro-cylindrical ion trap (μ-CIT) mass spectrometer arrays (r _o , z _o  ? 360 μm) is presented along with a strategy for incorporating integrated high-field emission electron sources for ionization. Simulations of the performance of μ-CITs were used to guide the geometrical design of the array, and process flows for microfabrication in silicon were developed and optimized for reduced capacitance. The microfabrication process is described in detail along with initial test results. Field emission sources were also fabricated in silicon using an edge-to-edge emission strategy to improve robustness in moderate vacuum environments. Simulations were also used to guide the design of the field emitter arrays. The microfabrication process is described and initial field emission data are presented. Strategies for further optimization and integration are also briefly discussed.     

Adsorbent Evaluation Based on Experimental Breakthrough Curves: Separation of p‐Xylene from C8 Isomers

The search for new adsorbents with enhanced capacity and selectivity, suitable for application on large‐scale simulated moving‐bed units for separation of p‐xylene, requires efficient, reliable, and fast adsorbent characterization methods for this specific separation. Fixed‐bed experiments were carried out under the conditions of the Parex process to evaluate a faujasite‐type zeolite as adsorbent for the separation of p‐xylene from its isomers in the proportions of the real Parex feed stream. The experimental breakthrough curves were used to evaluate the selected adsorbent in terms of nonselective and selective volumes, adsorption capacity, selectivity, and productivity, which can be applied to identify the feasible separation region for different operating conditions.     

Poly(n-butyl acrylate)/polystyrene interpenetrating polymer networks and related materials. III. Effect of grafting level and molecular weight in semi-2 IPNs

A series of poly(n-butyl acrylate)/polystyrene IPNs and semi-1 IPNs with deliberately controlled graft levels were synthesized via a urethane chemical coupling method. Also prepared were a series of semi-2 IPNs with the molecular weight of polymer II as the variable. The more highly grafted IPNs displayed poorly defined morphologies in which the domain structures were irregular and phase domain boundaries were characterized by fibrillar and interphase regions. A single glass transition peak was another feature of the more highly grafted IPNs. Polymer network II formed in the presence of linear polymer I results in morphologies dependent on the molecular weight of linear polymers. In the semi-2 IPNs, polymer I molecular weights below M_v = 20,000 caused polymer I to behave like a plasticizer or a diluent. The domain sizes of semi-2 IPNs agree with theoretical predictions developed by the present authors.