Linear Ramp Research Articles

Beneficial effects of time-restricted feeding on gentamicin cytotoxicity in mouse cochlea and vestibular organs.

ObjectiveTime restricted feeding (TRF), which is an intermittent fasting protocol, has been reported to decrease the toxicity and mortality rate associated with systemic gentamicin (GM) administration. The aim of this study is to evaluate the effect of TRF on GM‐induced vestibular and auditory function in mice.MethodsJapan Central Laboratory for Experimental Animals:Institute of Cancer Research (Jcl:ICR) mice were housed in a light–dark (LD) cycle (12:12) and were divided into three groups: (1) GM treatment at a dose of 220 mg/kg with TRF (feeding time: 8 h [9:00–17:00] during the light phase [7:00–19:00]) (GM + TRF group), (2) GM treatment at a dose of 220 mg/kg without TRF (GM group), and (3) saline injection with TRF (NS + TRF group). GM or saline was injected subcutaneously for 18 days (three courses of 5 days' injection + 2 days' rest, and an additional 3 days' injection). The auditory brainstem response (ABR) and vestibular evoked potential (VsEP) were tested after the treatments. The number of sensory hair cells in the cochlear organs and the vestibular organs were quantified using microscopic images.ResultsAll animals survived until the end of the experiment. One day after the last injection, GM + TRF mice showed significantly lower ABR thresholds at 4 kHz compared to GM mice, and there was no significant difference between the GM + TRF and NS + TRF groups. There was a significant difference of VsEP between GM and GM + TRF mice only in symmetric parabolic waves with linear acceleration and ramps waveform stimulation. GM + TRF mice showed significantly less outer and inner hair cell loss compared to GM mice. GM + TRF mice showed significantly less type II hair cell loss in the utricle and the ampulla compared to GM mice.ConclusionTRF with daytime feeding reduced GM cytotoxicity in the cochlea and vestibular organs of ICR mice.Level of EvidenceNA

Polyurethane Wood Adhesives Prepared from Modified Polysaccharides.

This study investigated the performance of polyurethane adhesives prepared with various combinations of wheat starch that had been modified by isophorone diisocyanate (MS), two polyol types (1,3-propanediol (PD) and glycerol (Gly)), native wheat starch (NS), and 4,4′-diphenylmethane diisocyanate (pMDI) at a NCO:OH weight ratio of 1:1. Two more adhesives were also synthesized with NS, PD, or Gly and pMDI blends and served as controls. The thermal behavior of the adhesives before and after the curing process, as well as their rheological performance and lap shear strength, were analyzed. Differential scanning calorimetry (DSC) showed a reduction in curing temperature and heat by adding MS. The thermal stability of the cured adhesives was slightly increased by MS addition. The viscosity of the adhesives that contained MS substantially increased at a linear ascendant ramp of shear, while the controls exhibited relatively low viscosity during the whole shear rate spectrum from 0.1 to 100 s−1. The tensile shear strength of wood veneers was also significantly increased by the incorporation of MS under both dry and wet measuring conditions. The maximum dry shear strength was obtained for the adhesive with Gly polyol and a higher content of MS and was comparable to the control adhesive with pMDI.

A Tail of Two Clumps

We present two axisymmetric simulations of a high velocity clump in a photoionized region: one for the case of a uniform, low density environment and a second one for the case of a clump first traveling within a high density medium and then emerging into a low density environment. We show that the second scenario results in the production of an axial tail of dense material with a linear velocity vs. position ramp (with zero velocity at the high/low density environment transition). This material comes from a confined bow shock (produced by the clump when it was within the dense cloud) that emerges into the low environmental density region.

Comparison of the Ramp and Step Incremental Exercise Test Protocols in Assessing the Maximal Fat Oxidation Rate in Youth Cyclists.

The incremental exercise test is the most common method in assessing the maximal fat oxidation (MFO) rate. The main aim of the study was to determine whether the progressive linear RAMP test can be used to assess the maximal fat oxidation rate along with the intensities that trigger its maximal (FATmax) and its minimal (FATmin) values. Our study comprised 57 young road cyclists who were tested in random order. Each of them was submitted to two incremental exercise tests on an electro-magnetically braked cycle-ergometer - STEP (50 W·3 min-1) and RAMP (~0.278 W·s-1) at a 7-day interval. A stoichiometric equation was used to calculate the fat oxidation rate, while the metabolic thresholds were defined by analyzing ventilation gases. The Student’s T-test, Bland-Altman plots and Pearson’s linear correlations were resorted to in the process of statistical analysis. No statistically significant MFO variances occurred between the tests (p = 0.12) and its rate amounted to 0.57 ± 0.15 g·min-1 and 0.53 ± 0.17 g·min-1 in the STEP and RAMP, respectively. No statistically significant variances in the absolute and relative (to maximal) values of oxygen uptake and heart rate were discerned at the FATmax and FATmin intensities. The RAMP test displayed very strong oxygen uptake correlations between the aerobic threshold and FATmax (r = 0.93, R2 = 0.87, p < 0.001) as well as the anaerobic threshold and FATmin (r = 0.88, R2 = 0.78, p < 0.001). Our results corroborate our hypothesis that the incremental RAMP test as well as the STEP test are reliable tools in assessing MFO, FATmax and FATmin intensities.

Computing the Capacitance of Electric Double-Layer Capacitors from Time-Domain and Frequency-Domain Measurements

In this study we show the relationship between capacitances of electric double-layer ca- pacitors (EDLC) computed from time-domain and frequency-domain measurements. EDLC are assumed to be linear time-invariant (LTI), causal and stable systems that map an input time-domain voltage signal v = [v0, v1, . , vN ] to an output time-domain charge q = [q0, q1, ..., qN ] = [f(v0), f(v1), ..., f(vN)]. When a voltage v0 is applied on an uncharged EDLC, the corresponding measured charge is q0 = c0v0 , where c0 is the value of the system impulse response at that instant. Subsequently, by considering the superposition principle in causal LTI systems, the input voltage v1 applied after a lapse of time ∆t with the impulse response of the device c0 results in a certain amount of charge c0v1 , but the preceding value of voltage v0 contributes also with another amount of charge c1v0 that depends on the current response of the device c1 at the instant ∆t. Thus the charge q1 accumulated on the device after ∆t is the sum c0v1 + c1v0 . Repeating this procedure for all discrete times tk = k∆t until N∆t (time duration of the applied signal) corresponds to the (discrete) convolution sum of the input voltage by the systems impulse response which results in the output charge: qτ = c ∗ v =Σj cτ− j vj . In continuous-time form, one can write: q(t) = ∫ c(t − τ )v(τ )dτ. The function c(t) can be viewed as the system-level, macroscopic capacitance of the device that takes into account (in a lumped form) all microscopic events of charge transport taking place inside the device. The convolution operation is carried out here in the time domain, but in the frequency domain, by virtue of the convolution theorem, it turns out to be the point-by-point multiplication Q = CV , where Q, C and V are the Fourier transforms of q, c and v respectively [1–3]. Thus, one cannot compute the capacitance of EDLCs by dividing charge-by-voltage both in the time-domain and frequency-domain as commonly done in the literature [3]. We will show examples using linear voltage ramp and constant current excitations [1,2].REFERENCES A. Allagui and M. E. Fouda, “Inverse problem of reconstructing the capacitance of electric double-layer capacitors,” Electrochim. Acta , 138848 (2021).A. Allagui, A. S. Elwakil, and M. E. Fouda, “Revisiting the time-domain and frequency-domain definitions of capacitance,” IEEE Trans. Electron Devices 68 (2021).A. Allagui, A. S. Elwakil, and H. Eleuch, “Highlighting a common confusion in the computation of capacitance of electrochemical energy storage devices,” J. Phys. Chem. C 125, 9591–9592 (2021).

Spectral Statistics of Non-Hermitian Matrices and Dissipative Quantum Chaos.

We propose a measure, which we call the dissipative spectral form factor (DSFF), to characterize the spectral statistics of non-Hermitian (and nonunitary) matrices. We show that DSFF successfully diagnoses dissipative quantum chaos and reveals correlations between real and imaginary parts of the complex eigenvalues up to arbitrary energy scale (and timescale). Specifically, we provide the exact solution of DSFF for the complex Ginibre ensemble (GinUE) and for a Poissonian random spectrum (Poisson) as minimal models of dissipative quantum chaotic and integrable systems, respectively. For dissipative quantum chaotic systems, we show that the DSFF exhibits an exact rotational symmetry in its complex time argument τ. Analogous to the spectral form factor (SFF) behavior for Gaussian unitary ensemble, the DSFF for GinUE shows a "dip-ramp-plateau" behavior in |τ|: the DSFF initially decreases, increases at intermediate timescales, and saturates after a generalized Heisenberg time, which scales as the inverse mean level spacing. Remarkably, for large matrix size, the "ramp" of the DSFF for GinUE increases quadratically in |τ|, in contrast to the linear ramp in the SFF for Hermitian ensembles. For dissipative quantum integrable systems, we show that the DSFF takes a constant value, except for a region in complex time whose size and behavior depend on the eigenvalue density. Numerically, we verify the above claims and additionally show that the DSFF for real and quaternion real Ginibre ensembles coincides with the GinUE behavior, except for a region in the complex time plane of measure zero in the limit of large matrix size. As a physical example, we consider the quantum kicked top model with dissipation and show that it falls under the Ginibre universality class and Poisson as the "kick" is switched on or off. Lastly, we study spectral statistics of ensembles of random classical stochastic matrices or Markov chains and show that these models again fall under the Ginibre universality class.

VELOCITY SEGREGATION IN A CLUMP-LIKE OUTFLOW WITH A NON-TOP HAT VELOCITY CROSS-SECTION

High velocity clumps joined to the outflow source by emission with a “Hubble law” ramp of linearly increasing radial velocity vs. distance are observed in some planetary nebulae and in some outflows in star formation regions. We propose a simple model in which a “clump” is ejected from a source over a period τ0, with a strong axis to edge velocity stratification. This non-top hat cross section results in the production of a highly curved working surface (initially being pushed by the ejected material, and later coasting along due to its inertia). From both analytic models and numerical simulations we find that this working surface has a linear velocity vs. position ramp, and therefore reproduces in a qualitative way the “Hubble law clumps” in planetary nebulae and outflows from young stars.

Multiobjective Dynamic Optimization for Optimal Load-Following of Natural Gas Combined Cycle Power Plants under Stress Constraints

With the increasing penetration of intermittent renewable energy sources into the electric grid, there is an associated need for conventional thermal power plants that are designed to operate at base-load conditions to cycle their load rapidly and frequently and operate under low-load conditions. Fast load-following operations lead to plant efficiency loss and the reduction of equipment life. A high-fidelity dynamic model of a natural gas combined cycle power plant, with rigorous equipment level submodels, is developed to capture the plant transient performance and the thermomechanical stress evolution at the high-pressure drum is computed to assess the drum life consumption. Stress evolution is modeled for the location of the edge at the drum/downcomer junction that experiences high circumferential stress amplitude during the fast load-following operation, leading to higher fatigue damage than locations considered under existing design standards. A dynamic optimization problem is solved for optimal load-following operation. Depending on the value of the stress constraint and the desired ramp rate, satisfying the desired stress constraint may be infeasible without relaxing the ramp rate. A multiobjective dynamic optimization problem is solved using a lexicographic approach that minimizes ramp rate relaxation and maximizes efficiency while satisfying stress constraints, avoiding spraying to saturation, and maintaining main steam and reheat steam temperature within bounds. It was observed that the optimal ramp rate can be highly nonlinear as opposed to the industry-standard linear ramp rate. Nonlinear optimal ramp-rate profiles not only help to avoid stress constraints that may be unavoidable by using the linear profile but also result in higher efficiency than the linear profile. The study shows that there is a strong tradeoff between the relaxation in the plant ramp rate and the time-average thermal efficiency.

Variability of cardiorespiratory interactions under different breathing patterns

The breathing dynamics often change in time and cause different variations in the cardiorespiratory interaction. There exist various breathing patterns, among them one critically important is the variability of the breathing frequency. We investigated the respiratory and the coupled cardiorespiratory system under controlled time-varying breathing patterns. Four breathing scenarios were used for this: spontaneous breathing, one where the subjects changed their breathing frequency according to linear ramp law, another according to a sine law and third according to an aperiodic predefined law. We introduced a framework of variability measures to trace and quantify the effect from the time-varying breathing perturbations. In particular, we studied intra-subject time-average variability, inter-subject subject-average variability and residual variability. These variability measures were estimated from the coupling strength and the similarity of coupling functions, for which we used methods specifically able to follow the time-evolving dynamics – the time–frequency wavelet transform and the adaptive dynamical Bayesian inference. The results demonstrated that the coupling and similarity were significantly greater in controlled, compared to free spontaneous breathing in many cases (p<0.0083). There were differences also among different controlled breathing regimes, and they appear both for intra-subject and inter-subject analysis. However, when the specific breathing perturbation is taken out, the results for the residual variability and the averaged coupling functions showed that the underlying interaction mechanisms remain invariant and not significantly different from spontaneous breathing (p>0.0083). This variability framework carries implications and can be applied more generally to other coupled oscillators and networks.

Quantum Kibble-Zurek mechanism: Kink correlations after a quench in the quantum Ising chain

The transverse field in the quantum Ising chain is linearly ramped from the para- to the ferromagnetic phase across the quantum critical point at a rate characterized by a quench time $\tau_Q$. We calculate a connected kink-kink correlator in the final state at zero transverse field. The correlator is a sum of two terms: a negative (anti-bunching) Gaussian that depends on the Kibble-Zurek (KZ) correlation length only and a positive term that depends on a second longer scale of length. The second length is made longer by dephasing of the state excited near the critical point during the following ramp across the ferromagnetic phase. This interpretation is corroborated by considering a linear ramp that is halted in the ferromagnetic phase for a finite waiting time and then continued at the same rate as before the halt. The extra time available for dephasing increases the second scale of length that asymptotically grows linearly with the waiting time. The dephasing also suppresses magnitude of the second term making it negligible for waiting times much longer than $\tau_Q$. The same dephasing can be obtained with a smooth ramp that slows down in the ferromagnetic phase. Assuming sufficient dephasing we obtain also higher order kink correlators and the ferromagnetic correlation function.

Inverse problem of reconstructing the capacitance of electric double-layer capacitors

The capacitance is a characteristic function of a capacitive energy storage device that is inaccessible to direct measurements, but can be estimated from input and output signals. Knowing that electric double-layer capacitors (EDLC) exhibit non-ideal capacitive behavior, their capacitance is commonly estimated from the time-domain definition of capacitance times voltage giving charge (using cyclic voltammetry and galvanostatic charge/discharge data), and in the same time from the same multiplicative relationship but in the frequency-domain (using impedance spectroscopy data), as if the capacitance is constant. The purpose of this study is to provide the recommended procedure to compute the capacitance of such types of devices from time-domain data starting from the definition of capacitance being the ratio of charge by voltage both defined in the frequency domain, which is consistent with the definition of conventional impedance. This turns the situation in the time domain to be an ill-posed inverse problem of convolution, wherein the to-be-deconvolved or reconstructed capacitance can be very sensitive to experimental errors. We show results of the procedure using (i) synthetic data generated using a fractional-order capacitor model of impedance Z(jω)=1/[(jω)αCα] and (ii) experimental data of a commercial aluminium electrolytic capacitor and an EDLC for the two cases of linear voltage ramp and constant-current step excitations.

Steady-periodic thermal stresses in an infinite hollow compound cylinder

Closed-form expressions for temperatures and elastic thermal stresses in an infinite hollow two-material compound cylinder subject to steady-periodic sinusoidal ambient temperatures are presented, taking finite heat transfer coefficients into account. The fluctuating temperatures and stresses in the cylinder are expressed in terms of complex-valued modified Bessel functions under the assumption of temperature-independent material properties. Transient thermal loads composed of linear temperature ramps are expressed as Fourier series, and studied by the presented method. In order to verify the analytic results, examples involving sinusoidal and linear loads are examined by finite element calculations.

Energy loss and jerk on the loop-the-loop

In its usual form, the loop-the-loop (LtL) problem involves a uniform solid sphere rolling from rest down a linear ramp that transitions into a circular loop. The task is to find the minimum height from which the ball must be released in order to roll completely around the loop without breaking contact. The answer, found using the conservation of mechanical energy and Newton's second law, is invariably less than the actual measured height. The difference, attributed to non-conservative forces, is consistently larger than the experimental uncertainty. To get a more detailed understanding of the effects of dissipative forces on the loop-the-loop, we made high speed video recordings of balls moving on the commercial LtL apparatus and used video analysis to study their motion in detail. We present our results along with a simple model to predict the motion of the ball on an LtL track taking energy losses into account. Calculations based on the model are in excellent agreement with our measurements.

Thermal Decomposition of Dimethyl Methylphosphonate on Size-Selected Clusters: A Comparative Study between Copper Metal and Cupric Oxide Clusters

Room temperature decomposition and thermal decomposition of dimethyl methylphosphonate (DMMP), a chemical warfare agent (CWA) simulant, on size-selected copper clusters have been studied via combined X-ray photoelectron spectroscopy (XPS) and temperature-programmed desorption (TPD). Cu100 and (CuO)80, which have the same nominal masses, were chosen to present a direct comparison between the reactivity of metallic copper and that of cupric oxide with DMMP. Room temperature XPS results have shown that most of the DMMP molecules decompose completely and reductively into atomic phosphorus on Cu100, while almost all the DMMP molecules are only dissociatively adsorbed on (CuO)80 as methyl methylphosphonate (MMP). XPS and TPD have been carried out to analyze the thermal decomposition of adsorbed DMMP by identifying the surface species after annealing to certain temperatures and the gaseous products evolved during linear temperature ramps, respectively. Methanol, formaldehyde, and methane are the three most significant gaseous products for DMMP decomposition on both Cu100 and (CuO)80. Methanol and formaldehyde, which evolve in the low temperature region, are believed to originate from surface methoxy species. Methanol, formaldehyde, and methane evolved in the high temperature region are related to further decomposition of the phosphorus-containing surface species. A set of methanol-probed TPD experiments have also been carried out, which suggest that methane evolution originates from the methyl group within DMMP instead of the surface methoxy species.

Modelling of a Torrefaction Process Using Thermal Model Object

So far, the torrefaction process has been merely discussed based on the physical and chemical characteristics of the final product, but the mechanism has not been yet pondered and investigated. Thus far, microwave torrefaction has been predominately used for thermal pre-treatment; therefore, a paradigm shift in the methodology has been introduced by using a Joule heating system. The article mainly focuses on the thermal engineering aspect of the torrefaction process. The densified black pine underwent thermal pre-treatment at a temperature of 523 K. The furnace used for torrefaction was initially improvised to carry out thermal degradation at quasi-static/dynamic conditions. A 3D PDE thermal model was developed to determine the numerical solution and temperature distribution across a black pine pellet. To compare the effect of the linear ramping profile, time-dependent as well as fixed Dirichlet conditions were applied to the proposed model. The mass distribution, duration of the torrefaction process, the effect of Nusselt and Reynold’s number of inert gas, and thermal history are some of the factors whose influence on the numerical solution was investigated. The simulation of thermal pre-treatment and its effect on the heat transfer characteristic was examined with help of a PDE thermal model, whereas the numerical solution of diffusion of the product of reactions was determined by solving the partial differential equations with the help of the discretisation method (PDEPE). The densification of black pine was performed in a ring die, whereas initial milling of biomass was carried out using a 1.5 mm sieve size. The system was found to have a homogeneous distribution in energy and temperature with time, whereas the amplitude of heat flux along the radial direction was reduced by 15% if the same pellet underwent torrefaction for a duration of 5 min in dynamic mode. Similarly, a 64.46% drop in amplitude of heat flux along the azimuth plane could be seen while performing torrefaction in a time-dependent thermal history. However, the relative amplitude of the heat flux at the centre of the pellet was estimated to be lowered by 98.41% along the vertical axis for heating a pellet in a quasi-static condition. The net change in the mass fraction of carbon dioxide across the boundary film was seen to be 40% higher than that of carbon monoxide. The rate of change of mass fraction of carbon monoxide across the boundary film was increased by 7–11% with the increase in torrefaction time. A 6.8% rise in the evaporation of water was noticed during the first half interval of torrefaction (from 5 min to 10 min). In the second half, from 10 min to 15 min, it was merely increased by 5.8%. A relative drop of 17.24% in water evaporation was estimated in the dynamic state of the system.

Phase retrieval using bidirectional interference.

In this paper we describe a phase retrieval algorithm using constraints given by diffraction patterns and phase difference obtained from bidirectional interference. Wave propagation and linear phase ramps are used to connect the recordings. At least three patterns are recorded and processed (two diffraction patterns and one interference pattern). The quality of the results can be improved when recording and processing more patterns. The method works well with non-sparse samples and short (few millimeter) recording distances. Simulations, comparisons with other methods, and experimental validations are presented.

Mass Ejection in Failed Supernovae: Equation of State and Neutrino Loss Dependence

Abstract A failed core-collapse supernova from a nonrotating progenitor can eject mass due to a weakening of gravity associated with neutrino emission from the protoneutron star. This mechanism yields observable transients and sets an upper limit on the mass of the black hole (BH) remnant. Previous global simulations of this mechanism have included neutrino losses parametrically, however, with direct implications for the ejecta mass and energy. Here we evolve the inner supernova core with a spherically symmetric, general-relativistic neutrino radiation-hydrodynamic code until BH formation. We then use the result in a Newtonian code that follows the response of the outer layers of the star to the change in gravity and resolves the surface pressure scale height. We find that the dense-matter equation of state (EOS) can introduce a factor of ∼2 variation in gravitational mass lost to neutrinos, with a stiff EOS matching previous parametric results and a soft EOS yielding lower ejecta masses and energies by a factor of several. This difference is caused primarily by the longer time to BH formation in stiffer EOSs. With a soft EOS, our red and yellow supergiant progenitors fail to unbind mass if hydrogen recombination energy is not included. Using a linear ramp in time for mass-energy lost to neutrinos (with suitable parameters) yields a stellar response within ∼10% of that obtained using the detailed history of neutrino losses. Our results imply quantitative but not qualitative modifications to previous predictions for shock breakout, plateau emission, and final BH masses from these events.

Investigating tertiary amine alkylation/benzylation kinetics with ramp-flow in a plug-flow reactor using in-line [formula omitted]H NMR spectroscopy

Quaternary ammonium salts are readily utilised within the chemical industry as, e.g., phase-transfer catalysts and anti-microbials, and their annual production capacity exceeds 1 million tonnes. These surfactants are conventionally produced in batch Menshutkin reactions from tertiary amines and an alkyl halide. With an incentive to move towards a more efficient continuous process, a better understanding of its kinetics is paramount. This paper explores a novel method to determine kinetic parameters, by introducing a linear flow ramp, combined with in-line Proton Nuclear Magnetic Resonance (1H NMR) analytics.Batch reactions between N,N-dimethyldecylamine (DMDA) and benzyl chloride (BnCl) have been performed in various solvents to examine solvent effects, and to validate the ramp-flow method. 1H NMR spectrometry showed that acetonitrile had the highest reaction rate, while methanol (MeOH) was deemed to have the best overall potential with regards to process intensification and product separation.Kinetic data was also obtained in a Plug-Flow Reactor (PFR) subjected to ramp-flow, under continuous measurement of the outlet stream composition. Between 20 and 70 °C, DMDA quaternisation rates were determined in MeOH, using both BnCl and methyl iodide reactants. This resulted in the successful elucidation of the respective Arrhenius parameters.Furthermore, the data from the ramp-flow method has been compared with the data obtained in batch, and in the PFR under conventional steady-state conditions. In general, all three methods were in good agreement, and it was concluded that the proposed method is valid. This method can thus provide a more efficient way to determine highly accurate kinetic parameters.

Thermofield dynamics: Quantum chaos versus decoherence

Quantum chaos imposes universal spectral signatures that govern the thermofield dynamics of a many-body system in isolation. The fidelity between the initial and time-evolving thermofield double states exhibits as a function of time a decay, dip, ramp and plateau. Sources of decoherence give rise to a nonunitary evolution and result in information loss. Energy dephasing gradually suppresses quantum noise fluctuations and the dip associated with spectral correlations. Decoherence further delays the appearance of the dip and shortens the span of the linear ramp associated with chaotic behavior. The interplay between signatures of quantum chaos and information loss is determined by the competition among the decoherence, dip and plateau characteristic times, as demoonstrated in the stochastic Sachdev-Ye-Kitaev model.

Competitive Online Convex Optimization With Switching Costs and Ramp Constraints

We investigate competitive online algorithms for online convex optimization (OCO) problems with linear in-stage costs, switching costs and ramp constraints. While OCO problems have been extensively studied in the literature, there are limited results on the corresponding online solutions that can attain small competitive ratios. We first develop a powerful computational framework that can compute an optimized competitive ratio based on the class of affine policies. Our computational framework can handle a fairly general class of costs and constraints. Compared with other competitive results in the literature, a key feature of our proposed approach is that it can handle scenarios where infeasibility may arise due to hard feasibility constraints. Second, we design a robustification procedure to produce an online algorithm that can attain good performance for both average-case and worst-case inputs. We conduct a case study on Network Functions Virtualization (NFV) orchestration and scaling to demonstrate the effectiveness of our proposed methods.