Measured Process Outputs Research Articles

Prevention validation and accounting platform: a framework for establishing accountability and performance measures of substance abuse prevention programs.

The field of substance abuse prevention has neither an overarching conceptual framework nor a set of shared terminologies for establishing the accountability and performance outcome measures of substance abuse prevention services rendered. Hence, there is a wide gap between what we currently have as data on one hand and information that are required to meet the performance goals and accountability measures set by the Government Performance and Results Act of 1993 on the other. The task before us is: How can we establish the accountability and performance measures of substance abuse prevention programs and transform the field of prevention into prevention science? The intent of this volume is to serve that purpose and accelerate the processes of this transformation by identifying the requisite components of the transformation (i.e., theory, methodology, convention on terms, and data) and by introducing an open forum called, Prevention Validation and Accounting (PREVA) Platform. The entire PREVA Platform (for short, the Platform) is designed as an analytic framework, which is formulated by a collectivity of common concepts, terminologies, accounting units, protocols for counting the units, data elements, and operationalizations of various constructs, and other summary measures intended to bring about an efficient and effective measurement of process input, program capacity, process output, performance outcome, and societal impact of substance abuse prevention programs. The measurement units and summary data elements are designed to be measured across time and across jurisdictions, i.e., from local to regional to state to national levels. In the Platform, the process input is captured by two dimensions of time and capital. Time is conceptualized in terms of service delivery time and time spent for research and development. Capital is measured by the monies expended for the delivery of program activities during a fiscal or reporting period. Program capacity is captured by fourteen measurement units, tapping into the dimensions of staff resources and community assets. Staff resources are, in turn, operationalized in terms of staff size, staff certification status, staff turnover rate, and the accreditation status of a provider agency. Community assets are operationalized by the number of community centers accessible to the funded agency, number of formalized teams or antidrug coalitions active in the catchment area, and other social/human services providers with whom the prevention agency has formalized networks. The totality of process output from all sources of program activities is reduced to eighteen classes of measures. These are operationalized by thirty-three summary measures. Some of these include: total count of events facilitated; total number of clients served; average number of clients served per event; clients served by single and multiple program sessions; classification of target population in terms of the severity of risk as defined by the Institute of Medicine; age groups and race/ethnicity of clients served; number of program participants retained by recurring programs; number of clients who have completed the program; penetration rates to the target population; client attrition rates; average referral rates per provider per time interval; referral success rates; and so on. All process output measures specified in the Platform are derived from two broad classes of events classified as either products or services. The collectivity of these measures is expected to present a cost-effective, parsimonious, yet comprehensive picture of the entire spectrum of the process output, i.e., "what came out of the program as program activities". For the measurement of performance outcomes, two types of data are incorporated into the Platform: outcome data from individuals and the behavior (or performance) of social indicators from aggregated data bases. Individual data are used to evaluate the outcome of substance abuse programs

Nonlinear Observer Design for Process Monitoring

The present work proposes a systematic nonlinear observer design framework for process monitoring. The objective is to accurately monitor key process variables associated with process safety or product quality, by designing a model-based nonlinear observer that directly utilizes the available information coming from continuous-time on-line process output measurements. The nonlinear observer possesses a state-dependent gain which is computed from the solution of a system of first-order linear partial differential equations (PDEs). Depending on whether the process operates in continuous or batch mode, a different mathematical treatment and solution scheme for the system of PDEs is needed. Within the proposed design framework, both full-order and reduced-order observers are studied. Finally, the performance of the proposed observer is evaluated in two chemical-engineering examples: (1) a catalytic batch reactor where the objective is to accurately estimate the catalyst activity and (2) an exothermic CSTR that exhibits steady-state multiplicity and the heat released by the reaction needs to be closely monitored.

A modified index for control performance assessment

A control performance index which relates actual output variance to minimal achievable variance was first presented by Harris [T.J. Harris, Assessment of control loop performance, The Canadian Journal of Chemical Engineering 67 (1989) 856–861]. In the present paper, a modification of that index is applied. Instead of comparing to minimum variance control — which corresponds to placing all closed-loop poles in the origin — one pole is placed using either control design guidelines or additionally available process knowledge. Some possible choices of the free closed-loop pole are discussed. Based on this modified closed-loop system, a performance index can be calculated which relates actual output variance to the minimum achievable variance using the modified closed-loop system. It is shown that for the calculation of this index no other information apart from the measured process output and the time-delay is necessary. Furthermore, a method is presented which allows the assessment of the accuracy of the calculated index. The original Harris index and the proposed modified index are computed using data from industrial processes.

On-line process identification and PID controller autotuning

We propose a new and simple on-line process identification method for the automatic tuning of the PID controller. It does not require a special type of test signal generators such as relay or P controller only if the signals are persistently exciting. That is, a user can choose arbitrary signal generators such as relay, a P controller, the controller itself, pulse signal and step signal generator because it needs only the measured process output and the controller output. It can incorporate nonlinearities due to actuator saturation or manual mode operation during identification work and shows a good robustness to measurement noises, nonlinearity of the process and disturbances. The proposed autotuner combined with the identification method and tuning rule using a model reduction shows good control properties compared with previous autotuning methods.

On-line process identification and automatic tuning method for PID controllers

We propose a new and simple on-line process identification method for the automatic tuning of the PID controller. A least-squares method is used to obtain the model from measured process 0outputs and inputs without any structural information. Thus, any kind of test signal generators such as a relay, a P controller, the controller itself, pulse and step signal generators, etc can be utilized only if they contain sufficient information because it requires only the measured process output and the controller output. General time-invariant linear systems including open-loop stable, nonminimum phase, unstable and integrating processes can be identified effectively by the proposed identification method. We then use a model reduction method to tune the PID controller using usual tuning rules based on the first- or second-order plus time-delay model.

Scarce Data Operating Conditions: Process Model Identification

Abstract In this paper, properties of Pseudo LS algorithms are analysed and some applications are discussed. Identification of discrete-time models of a continuous-time system subject to a missing data unconventional sampling pattern is considered, i.e., process output measurements are obtained with a lower availability rate than for process input ones. The suitability of this approach for fault detection applications is also shown.

Digital Model Parameter Estimation with Missing Data

In this paper, a modified least squares parameter estimation algorithm dealing with scarce data is presented to tackle the problem of tracking a reduced number of parameters in a continuous-time system. The process output measurements may be obtained in various ways, always assuming for them a lower availability than for process input ones. Particular cases here considered are: periodic sampling at a lower rate, random sampling, and a set of simple on-off sensors that trigger an interrupt when a certain system element reaches a precisely known position or voltage (value-driven sensors). This approach is to be shown mainly interesting for fault detection applications.

Simulation of quantitative fault diagnosis in backfill hydrocyclones

A quantitative fault diagnosis system for backfill hydrocyclones using modelling and parameter estimation was developed. The system was evaluated using simulation, thereby allowing arbitrary process variable fault levels and combinations of faults to be introduced into the system, as well as controlled levels of random noise in the process output measurements. The fault diagnosis system determines that point in the search space that minimises the difference between the measured process outputs and the process outputs predicted by previously determined process models by simultaneously adjusting the model parameters. At this minimum, estimates of the unmeasurable process variables are provided by the model parameters. A gradual increase in the feed −75 micron fraction was considered, as well as a simultaneous increase in both the vortex finder and spigot diameters. Faults could clearly be identified by observing the trends in inferred variable values over time. Subjective decisions as to the level of incipiency of faults can be reduced by performing quantitative alarm analysis on these values. Both non-parametric and parametric alarm analysis was evaluated and it was found that a combination of these methods would enable timeous detection of both gradual and abrupt faults. The use of quantitative fault diagnosis to detect incipient faults in process parameters could be of great value in maintaining production levels, planning maintenance and optimising system performance.

Identification of dynamic models for the Shell benchmark problem

Dynamic process and disturbance models are identified for the 2×2 distillation column described in the summary paper by Cott. Process models are identified using both classical least squares and the newer AUDI identification algorithm developed at the University of Alberta. The identified models are validated and gain-adjusted in the frequency domain by comparing the magnitude spectrum of the models, | G (e −iω )|, versus the spectrum obtained by fast Fourier transformation of the original plant I/O time series. The noise/disturbance models are identified using standard time series tools and all models are verified by showing that the filtered residuals (measured process output minus model based estimate of the output) are essentially white noise. The estimated SISO models are all first-order and very close to the actual process whether compared on the basis of step-response coefficients, model parameters or frequency domain plots. The control problem is so simple that two standard PID feedback loops with one-way (lead/lag) interaction compensation produce results that are essentially identical to a 2×2 multivariable, model-based (GPC) controller.

Soft-sensors for process estimation and inferential control

Abstract This paper presents two adaptive estimators (software based sensors or ‘soft-sensors’) for inferring process outputs that are subject to large measurement delays, from other (secondary) outputs which may be sampled more rapidly. In other words, these estimators utilize plant data sampled at different rates. The parameters of the estimators can be continuously estimated and updated on-line, thus enabling the tracking of slow variations in process characteristics. The estimators employ either an input-output or a state space process description as the starting points for algorithm synthesis. In contrast to mechanistic model-based estimators such as Kalman filters, the proposed adaptive techniques require minimal design effort. The key contribution of the paper is thus the formulation of applications independent, adaptive multi-rate algorithms, which can provide accurate estimates of infrequently measured process outputs, from other more rapidly sampled secondary outputs. Theoretical developments are supported by results of recent applications to a variety of industrial scale processes: estimation of biomass concentration in an industrial mycelial fermentation; top product composition of a large industrial distillation tower and melt flow index on an industrial polymerization reactor. Measurements from established instruments such as off-gas carbon dioxide in the fermenter; overheads temperature in the distillation column and hydrogen concentration in the reactor were used as the secondary variables for the respective processes. The range of the applications is an indication of the utility of the techniques. A significant improvement in overall process control performance is also possible when estimated plant outputs, rather than the infrequently obtained measurements, are used for feedback control. This is demonstrated by non-linear simulation studies.

Adaptive inferential control

An estimation technique for inferring infrequently measured process outputs, using other more rapidly sampled secondary outputs, is proposed. The estimator has a general structure and its parameters can be estimated online. In contrast to other estimation techniques in the literature, the proposed adaptive method requires minimal design effort and does not result in static estimation error. The problem of secondary output selection is therefore alleviated. Moreover, the estimator can track slow variations in both process and disturbance characteristics. Control schemes using the estimated values of the output are shown to yield considerably better performance than corresponding schemes which employ infrequently sampled values of the primary output.

Optimization of Discrete Additions of Substrate During the Batch Biosynthesis of Antibiotics

A case history is described in which a practical systems engineering approach was applied to the industrial biosynthetic production of antibiotics, with the aim to develop a better operating policy for a single fermentor. The control variables to be optimized are: the sequence of discrete additions of substrate, the initial concentration of substrate and the duration of the batch (batch time). Three different criteria were considered, namely batch profit, production rate and substrate utilization efficiency. A relatively simple (formally a chemical-kine-tic) model is shown, giving good agreement with the measured process outputs. For optimization, a simulation study and the Quasi Newton method (with simple bounds on variables) were combined. From the results obtained, the following practical recommendations are given: 1. the initial concentration of substrate should be remarkably increased, 2. a falling sequence for substrate additions should be introduced, 3. for the first two criteria, higher substrate addition rate is desirable, whereas for the third criterion substrate additions later during the batch are to be canceled, 4. batch time is generally not a critical variable; eventually it can be optimized by an univariable search. In the light of these conclusions, further research of substrate inhibition in this process is highly desirable.