Search

During recent decades, planning defense systems have evolved into capability-based planning (CBP) processes. This paper seeks to answer two questions: firstly, how to express a complex, real-world capability requirement; and secondly, how to assess if a system with interacting elements fulfills this requirement. We propose that both a capability need and the solution fulfilling it are expressed with a consistent set of models in a traceable manner. The models integrate current capability models, specific to planning level and capability viewpoint, with systems thinking approach. Our conceptual model defines the defense system in its environment, our data model defines and organizes the CBP terms, and our class diagram defines the CBP planning elements. We illustrate the approach by giving an example of capability parametrization and compare it both with the DODAF capability view and with the generic CBP process. Our data model describes how capabilities are degraded in action and extends the approach toward capability dynamics. The quantitative capability definition aims to support efforts to solve for real world interacting subsystems that combined implement the required capability. publishedVersion

The thickness profile of the tubular bubble in blown film extrusion is deduced from delayed measurements made downstream in the flatted web consisting of the bubble. Scanner moving across the web measures only a portion of it at a time and the measurement results are matched to the bubble profile at actuators controlling the thickness. The delayed profile is predicted to current time and used for LQR feedback control. The control architecture is tested in a simulation of the process. acceptedVersion

Quantization noise is present in many real-time applications due to the resolution of analog-to-digital conversions. This can lead to error in policies that are learned by model-free Q-learning. A method for quantization error reduction for Q-learning algorithms is developed using the sample time and an exploration noise that is added to the control input. The method is illustrated with discrete-time policy and value iteration algorithms using both a simulated environment and a real-time physical system. publishedVersion

The cross-directional (CD) basis weight control on paper machines is improved by optimizing the path of the scanning measurement. The optimal path results from an LQG problem and depends on how the uncertainty of the present estimate of the basis weight and the intensity of process noise vary in CD. These factors are assessed by how accurately the CD basis weight estimate predicts the measured optical transmittance with a linear adaptive model on synchronized basis weight and transmittance data. Simulations on optimized scanner path in disturbance scenarios are presented, and the practical implementation of scanner control is discussed.

Forest resource data is important in targeting the forestry operations, and it is in the hearth of the precision forestry concept. The forest resource data can be produced with many techniques, and the number of existing forest data sources has increased during the years. In addition to the forest resource data, other data describing the circumstances of the forest site, such as trafficability and weather conditions, are available. In Finland, a forest data platform gathers the data sources under a single service for easier implementation of the precision forestry applications. This data is useful in operations planning, but it also describes the conditions that prevail when the forest machine arrives to the forest site. This study proposes data fusion between fieldbus time series of the forest machine and the forest data. The fused dataset enables explorative statistical analysis for examining the relationship between the machine performance and the forest attributes and provides data for building predictive models between the two. The presented methods are applied into a dataset generated from a field test data. The results show that some fieldbus time series features are predictable from forest attributes with $$R^{2}$$R2 value over 0.80, and clustering methods help in interpreting the machine behavior in different environments. In addition, an idea for generating a new forest data source to the forest data platform based on the fusion is discussed.

In mechanized logging operations, interactions between the forest machines and their operators, forest resources and environmental conditions are multifold and not easily detected. However, increased computational resources and sensing capabilities of the forest machines together with extensive forest inventory data enable modeling of such relationships, leading eventually to better planning of the operations, better assistance for the forest machine operators, and increased efficiency of timber harvesting. In this study, both forest machine fieldbus data and forest inventory data were acquired extensively. The forest inventory data, acquired nationwide, was clustered to categorize general tree and soil types in Finland. The found forest categories were applied when the harvester fieldbus data, collected from the forest operations in the North Karelia region with two similar harvesters, was analyzed. When the performance of the machine and the operator, namely the fuel consumption and log production, is studied individually for each forest cluster, the impact of working environment no longer masks the causes based on the machine or the operator, thus making the observations from separate forest locations comparable. The study observed statistically significant differences in fuel consumption between the most general tree and soil clusters as well as between the harvester-operator units. The modeling approach applied, based on multivariate linear regression, finds such reasons for the differences that have clear interpretation from machine setup or operator working style perspective, and thus offers a feasible method for assisting the operators in improving their working practices and thus the overall performance specifically at forest of given type.

At present, the tissue paper manufacturing is mostly based on the dry crepe technology. During the last decade, the manufacturers have introduced new tissue machines concepts that increase the softness, bulk, and absorption capacity. Such machines produce a strong regular three-dimensional (3D) structure to the sheet before the Yankee cylinder. At present, the quality of the 3D structure is not evaluated, or it is evaluated only subjectively at the mill. This is mostly because of the difficulties to separate reliably the regular 3D pattern from other variations. This paper introduces a frequency analysis based method which separates the surface profile variances in tissue paper to the creping, to the regular 3D pattern and to the residual variation. The 3D surface profiles and their variances were determined online with the photometric stereo method. We show that the introduced analysis method evaluates the variance portions reliably and the results are consistent with the visual perception of the 3D surfaces. In one particular product, the regular 3D pattern explains 74 % of total surface variance; the creping explains 10 % and residual variations 16 %. Furthermore, the creping and residual variances are quite stable over time whereas the variance of the regular 3D pattern fluctuates significantly.

Network topology specifies the interconnections of nodes and is essential in defining the qualitative network behaviour which is known to be universal with the similar phenomena appearing in many complex networks in diverse application fields. Topology information may be uncertain, or several pieces of inconsistent topology information may exist. This paper studies a method for estimating the network topology directly from node data, and is motivated by mobile telecommunications networks (MTNs). Mutual information based dependency measure is first used to quantify the statistical node dependencies, and the topology estimate is then constructed with multidimensional scaling and distance thresholding. The topology estimate defines the graph structure of a Markov random field (MRF) model, and after model parameter identification, the MRF model can then be used e.g. in analyzing the effect of disturbances to the overall network state of MTN. The method is evaluated with MCMC generated data and is found to work in qualitative network behaviour situations that are practical from the application perspective of MTNs. With the same data, the method yields at least as good results as a typical constrained-based graph estimation method.

We consider the problem of an agent traversing a directed graph with the objective of maximizing the probability of reaching a goal node before a given deadline. Only the probability of the travel times of edges is known to the agent. The agent must balance between traversal actions towards the goal, and delays due to actions improving information about graph edge travel times. We describe the relationship of the problem to the more general partially observable Markov decision process. Further, we show that if edge travel times are independent and the underlying directed graph is acyclic, a closed loop solution can be computed. The solution specifies whether to execute a traversal or information-gathering action as a function of the current node, the time remaining until the deadline, and the information about edge travel times. We present results from two case studies, quantifying the usefulness of information-gathering as opposed to applying only traversal actions.

Mobile multi-machine environments consist of varying types of objects, either static or dynamic with the state known exactly or with some uncertainty. Sensors observe the environment from different positions and views, such as horizontally from the top of a mobile machine or vertically downward from an external observing machine. In this paper, we propose a simple grid-based framework for representing the information of varying types of objects in a 2-D environment and Bayesian methods for updating this information through observations and prediction models. This information about the current and near-future state of the environment is called situation awareness (SA). SA information can be utilized as the basis of the operator assistance system for enhancing the safety and efficiency of manual and semiautonomous multi-machine work environments. SA information can also be utilized in task planning of autonomous machines, via creating a dynamic costmap for path planning or entropy map for planning the optimal use of sensory systems. This paper aims at real-time mobile multi-machine environments; hence, the SA framework is kept simple for computational feasibility, but is also general enough to be applicable in other environments as well. Discussion on discretization errors and computational complexity are also covered.

Automatically collected forest environment data is essential when developing more accurate and efficient forestry operations. Reliable position-based data enables efficient wood procurement operations and helps to avoid damage to the forest floor. A modern forestry vehicle with extensive sensing capabilities could measure environmental parameters, such as soil type, topography or weather conditions, while carrying out other wood procurement tasks. Furthermore, methods to improve positioning accuracy are also called for when the positioning is based on global satellite navigation systems (GNSS), whose signals are often blocked by the forest canopy. In this article, data is collected automatically with two of Microsoft’s Kinect v2 time-of-flight sensors during field tests in a forest environment in Southern Finland. The aim of the article is to propose methods which will improve positioning accuracy by enabling the movement of the forwarder to be detected and also to provide reliable measurements of...

We study the cooperation of an unmanned ground vehicle (UGV) and micro air vehicle (MAV) in a path planning task. The UGV requests the MAV to execute observation missions which provide stereo image data on areas that the UGVs sensors cannot observe. The problem is formulated as a partially observable decision making problem. The solution is a decision policy that determines conditional on current information whether the UGV should move along a certain route or request an observation mission. A case example for a real-world demonstration of the cooperative path planning task is described. We discuss the features of single-vehicle control systems and propose a multi-vehicle communication framework that is tolerant against communication breaks.

Continuous-state partially observable Markov decision processes (POMDPs) are an intuitive choice of representation for many stochastic planning problems with a hidden state. We consider a continuous-state POMDPs with finite action and observation spaces, where the POMDP is parametrised by weighted sums of Gaussians, or Gaussian mixture models (GMMs). In particular, we study the problem of optimising the selection of measurement channel in such a framework. A new error bound for a point-based value iteration algorithm is derived, and a method for constructing a subset of belief states that attempts to reduce the error bound is implemented. In the experiments, applying continuous-state POMDPs for optimal selection of the measurement channel is demonstrated, and the performance of three GMM simplification methods is compared. Convergence of a point-based value iteration algorithm is investigated by considering various metrics for the obtained control policies.

Sequential decision making under uncertainty is studied in a mixed observability domain. The goal is to maximize the amount of information obtained on a partially observable stochastic process under constraints imposed by a fully observable internal state. An upper bound for the optimal value function is derived by relaxing constraints. We identify conditions under which the relaxed problem is a multi-armed bandit whose optimal policy is easily computable. The upper bound is applied to prune the search space in the original problem, and the effect on solution quality is assessed via simulation experiments. Empirical results show effective pruning of the search space in a target monitoring domain., Comment: 6 pages, 2 figures

This work presents a nonlinear optimization and scheduling approach applied to a copper production plant. The solution maximizes smelting furnace production and provides valid converting schedules by simulating the evolution of the process over the optimization horizon. The production process is briefly described and the main models used to predict and calculate furnace and converter parameters are detailed. Though the solution is concentrated on the main elements, copper and iron, the optimization framework enables easy future augmentation with more complex models. A schedule optimization case is presented.

The key operation in tissue production is the creping at the Yankee cylinder. The tissue sheet is adhered to a Yankee cylinder and then detached from the surface with a blade. As a result a strong microstructure - crepe folds - is generated on the web. In this work the structure of creping pattern in CD and MD of tissue paper was studied. The samples available in this work consisted of bath tissue and towel grades. The significant difference between the samples was in the number of crepe folds. The tissue sheets were analyzed with an imaging system. The variation of crepe folds in images was studied in spatial frequency space through 2D Welch spectrum. It was noticed that the difference in creping pattern structure between the samples can be seen clearly in spectra. The creping pattern was simulated based on the creping pattern variation found from the Welch spectra of tissue paper images. The simulation was based on a set of sinusoidal terms representing several orientations and amplitudes. The amplitudes of sinusoidal terms located in regular grid in spectra were approximated with 2D normal distribution. Thus, the structure of creping pattern can be represented with three parameters originating from the deviations and location of 2D normal distribution. The creping pattern reconstructed with simulations contains the relevant features in the creping pattern of true tissue images.

Accurate knowledge of the input impedance of a radio-frequency identification (RFID) integrated circuit (IC) at its wake-up power is valuable as it enables the design of a performance-optimized tag for a specific IC. However, since the IC impedance is power dependent, few methods exist to measure it without advanced equipment. We propose and demonstrate a wireless method, based on electromagnetic simulation and threshold power measurement, applicable to fully assembled RFID tags, to determine the mounted IC's input impedance in the absorbing state, including any parasitics arising from the packaging and the antenna-IC connection. The proposed method can be extended to measure the IC's input impedance in the modulating state as well.

Prediction of overall visual quality based on instrumental measurements is a challenging task. Despite the several proposed models and methods, there exists a gap between the instrumental measurements of print and human visual assessment of natural images. In this work, a computational model for representing and quantifying the overall visual quality of prints is proposed. The computed overall quality should correspond to the human visual quality perception when viewing the printed images. The proposed model is a Bayesian network which connects the objective instrumental measurements to the subjective opinion distribution of human observers. This relationship can be used to score printed images, and additionally, to computationally study the connections of the attributes. A novel graphical learning approach using an iterative evolve-estimate-simulate loop learning the quality model based on psychometric data and instrumental measurements is suggested. The network structure is optimised by applying evolutionary computation (evolve). The estimation of the Bayesian network parameters is within the evolutionary loop. In this loop, the maximum likelihood approach is used (estimate). The stochastic learning process is guided by priors devised from the psychometric subjective experiments (performance through simulation). The model reveals and represents the explanatory factors between its elements providing insight to the psychophysical phenomenon of how observers perceive visual quality and which measurable entities affect the quality perception. By using true data, the design choices are demonstrated. It is also shown that the best-performing network establishes a clear and intuitively correct structure between the objective measurements and psychometric data.

Measurement systems produce information to support decision making. The added value of measurements is determined by how much the provided information improves decision making as uncertainty about decision consequences is reduced. We discuss through an example the value generation by measurement information in the context statistical decision theory (SDT). This example, a grade change decision, outlines the principle for determining added value in decision making. We discuss separately two value generation mechanisms: the direct value generation by reduced uncertainty on consequences and the indirect value generation through accumulating knowledge about the consequence model parameters. The first mechanism is solvable in analytic form and this allows comparing formulations of SDT in the analysis. We outline the analysis of the indirect value generation and show that even with this simple decision task the analysis is rather complex. We propose that the added-value analysis is the key to optimal design of measurement information systems.

We shall present approaches to optimize the design of a measurement system and to schedule dynamically a versatile measurement resource. The analysis is based on expressing the system management task as a dynamic programming problem in which the system state is partially observable. We shall review the well-known linear-quadratic-Gaussian case, and discuss and give example solutions of discrete state systems with approximate dynamic programming methods to solve such problems in practice. Design problem is then studied by assessing the long-term performance of control and how it depends on properties of measurements. A simple design example is presented for a discrete state system.

Operational decisions are influenced not only by the data and models available to the decision maker but also by the uncertainty in the data and in model-based predictions about the impacts of decision makers' actions. In non-linear systems the potential actions may have widely differing uncertainty associated with them. Then the decision maker must take an attitude towards risk and balance that against the expectation value of performance. In stochastic optimization, methods to deal with uncertainty have been developed. However, these methods have not been widely used in decision making about operating industrial processes. In this article, we first present a short summary of decision making under uncertainty and then suggest that the mathematical structure of stochastic optimization serves as a model for the architecture of future operational decision support systems. We demonstrate this framework by analysing four idealized operational decision cases, which are closely related to practical daily decisi...

A bistable system with a desirable and an undesirable state exhibits transitions between the states because of stochastic actions on the system. Selecting the operational point of a bistable system is a tradeoff between expected performance of the system and risk of transition to an undesired state. Hence control of a bistable system must consider the probability density function of the state. Measurement about state reduces state uncertainty according to the Bayesian rule. The author analyses how the measurement frequency and uncertainty affect the tradeoff between performance and risk, and thus seek for optimal measurement design. The author discretises the time by Euler integrating the stochastic differential equation describing the bistable system over an interval much shorter than any deterministic time scale, and discretise state to describe the system as a Markov chain. Measurements are made infrequently - every 50-100 time steps - and their uncertainty is Gaussian. The tradeoff problem is formulated by maximising system performance subject to constraint on the probability of undesired transition. The tradeoff is then obtained by varying the constraining probability. Similar tradeoffs are obtained either by frequent but uncertain and less frequent but more accurate measurements. Our analysis is motivated by the control of bioreactors.

In pulp and paper the modelling and simulation of the pulp production processes was the first major application. Since then several other applications have been realised though papermaking has never been on the lead concerning the use of computer-based modelling and simulation. The complex nature of the materials is one of the most demanding challenges and the biggest hurdle for any electronic description of the papermaking process or paper itself. So other industries took the lead and it is now for the paper industry to learn from these industries what can be done with the help of computers to control or optimise processes or to design new grades.The application of results gained with the use of modelling and simulation techniques in pulp and paper manufacturing processes has helped the industry to, for example, reduce emissions and increase the productivity and cost-efficiency of the processes. Still there are many important tasks open. A better understanding of the mechanisms of the processes and their control loops has to be achieved in order to further improve the paper quality, stabilise the wet-end chemistry, and enhance the runnability. Important features to be developed are new process designs, efficient process monitoring and systems that offer decision support during operation.In this paper a review of the state of the art on modelling and simulation in the pulp and paper industry is presented as well as further research needs.

Operational decisions are influenced not only by the data and models available for the decision maker but also by the uncertainty within them. The decision maker must take an attitude towards risk and balance that against the expectation value of performance. This paper presents methods to manage rationally operational decision making tasks under uncertainty with an example taken from the paper industry, the scheduling of thermo-mechanical pulp (TMP) production with uncertainty in production cost and demand. The stochastic problem is analytically transformed into a deterministic optimization task, and the probability distributions of the stochastic variables are calculated directly rather than with Monte Carlo simulations. The objective is formulated in two alternative ways: either the decision maker's attitude towards risk is modeled with exponential utility function or the expected cost is minimized subject to an additional constraint for probability of exceeding user specified maximum cost allowed. The role of the constraints in a dynamic, stochastic case is also discussed.

Mobile multimachine work environments, in general, consist of varying types of machines driving across a site to complete a moving or manipulation task. As the view from the machines over the environment can be limited due to the structure of the machines, the environment, and the moved/manipulated items, there is a risk of collision between machines. In this paper, we propose a situation awareness system aimed as a driver/operator assistive system to enhance the safety and efficiency of multimachine work environments. The system consists of the pose estimation of the machines, the M2M communication based on IEEE 802.11p, the future pose prediction of the machines, and a graphical user interface. The system is designed to be general enough to be applicable in diverse work environments, which is most easily implemented as a retrofit for existing machines. In this paper, the implementation of the system presented is a proof of concept, and the focus is on how the overall system works in a real harbor environment during operation, ultimately aiming for the collision avoidance of harbor machines. The field tests show promising results, particularly regarding the applicability of the M2M communication technology in a very challenging and uncertain harbor environment.

We address the problem of controlling a mobile robot to explore a partially known environment. The robot's objective is the maximization of the amount of information collected about the environment. We formulate the problem as a partially observable Markov decision process (POMDP) with an information-theoretic objective function, and solve it applying forward simulation algorithms with an open-loop approximation. We present a new sample-based approximation for mutual information useful in mobile robotics. The approximation can be seamlessly integrated with forward simulation planning algorithms. We investigate the usefulness of POMDP based planning for exploration, and to alleviate some of its weaknesses propose a combination with frontier based exploration. Experimental results in simulated and real environments show that, depending on the environment, applying POMDP based planning for exploration can improve performance over frontier exploration., Comment: 19 pages, 11 figures in Robotics and Autonomous Systems (2016)

Dynamic systems, such as industrial production processes, are measured indirectly with fast on-line measurements. Occasionally, more direct and accurate measurements are made in laboratory to validate the on-line measurements. We discuss the optimal dynamic estimation of the parameters of characteristic curves under the assumption that their validity deteriorates through a stochastic process, describable through a stochastic differential equation. We provide the general formalism and an easy-to-implement algorithm for Gaussian distributions of measurement errors and random-walk deterioration of parameter validity. Furthermore, we analyze the dynamic estimation of the measurand given the earlier information about it deteriorates through (another) stochastic differential equation, and show the effect of uncertainty in the parameters of characteristic curve.

The production processes are becoming increasingly complex and the responsibilities of the operators and the engineers are widening. Tools to manage this complexity under dynamic conditions are needed. The dynamic optimization integrated to operator and engineer decision support has a high potential in everyday use. As a result of an EU Commission funded research project a toolset for supporting continuous decision making has been developed. This paper presents a case study, where this toolset is used to reduce grade change time on a paper machine.

acceptedVersion

This paper discusses data-based operating windows as a tool for process management and development. In particular identification of the operating window and its uncertainty are analyzed. The operating window is determined by maximizing either the mutual information (static) or entropy transfer (dynamic). An industrial example shows that the entropy of the indicator variable is reduced to half by an operating window specified with only few variables, selected amongst over 3000 candidates. Test model based simulations suggest that such few-variable operating windows can be reliably identified from datasets having lengths of a few thousand observations.

A control strategy for basis weight and filler content of paper is presented. The largest cross effect between basis weight and filler content is avoided by using fiber and filler weights for control. The effect of broke filler in thick stock is compensated for by feed-forward control. A circulation model with retention as a parameter is used as a Smith predictor model. The controllers for both fiber and filler are ordinary PI controllers.

This paper examines operational optimization problems of processes including binary stochastic disturbances. Such disturbances can move the operating point suddenly and increase the probability of some undesired events. In such cases, the operational optimization over a time horizon may lead to infeasibility further down in the horizon and hence prevent finding the optimal operational actions at present time. In this paper, we present a strategy for relaxing the constraints of such optimization problems. We illustrate the method by an industrial oriented case study from papermaking.

This paper presents an optimization strategy for pulp/water flow management in a paper production system. The problem is formulated at two levels - the upper/design level concerning optimal volumes of the pulp/water tower and the lower/operational level concerning how to control the system optimally with given tower volumes. At both levels of there are multiple and conflicting objectives.

This paper presents an optimization strategy for the design and operation of a broke management system in a papermaking process. A stochastic model based on a two-state Markov process is presented for the broke system and a multiobjective and bi-level stochastic optimization model is developed featuring (i) a multiobjective operational subproblem for the optimization of the broke dosage and (ii) a multiobjective design problem formulation. An efficient optimization strategy is proposed for the operational subproblem along with a simulation based Pareto optimal solution for the design problem, and illustrated with a detailed case study.

This paper presents an optimization strategy for a broke system design and operation in papermaking process. A multiobjective stochastic optimization model is presented featuring (i) a stochastic two-state Markov process based submodel for the broke tower, (ii) an operational submodel for the optimization of the broke dosage, and (iii) a multiobjective design problem. An efficient optimization strategy is also proposed involving a quadratic optimization step for the operational subproblem and an effective multiobjective design optimization step.

This paper discusses time series based quality analysis in mill-wide systems. A general purpose data source interface which integrates all data sources is discussed. Simulation is considered as an additional data source. The automatic derivation of simulators from process structures is analyzed and practical solution based on MATLAB/SIMULINK software is presented. The mill-wide implementation issues are discussed.

Printability of paper is a multidimensional concept consisting of runnability and print quality. The lack of definition for print quality makes printability rather ambiguous. In this paper we discuss the measurability of the high-level printability and print quality and present an interpretation of the measurements of low-level print quality related characteristics as information channels about high-level concepts. This enables the construction of a layered model in which the information provided by measurements is propagated to infer about higher level concepts of quality, eventually supporting decision making. As an example, we present a case study of low-level print quality defects caused by the small-scale measurable properties of the paper, and consider the meaning of this process from the viewpoint of print quality analysis. publishedVersion

We outline the needs for diagnostics in the paper industry and describe the state of the art. We have developed two next-generation diagnostic systems: WEDGE for the diagnostics of temporal variations, and FLEXPRO for spatial (cross-direction) variations. Both these systems combine powerful signal analysis tools with an intelligent user interface. Thus our systems can be used on a daily basis at paper mills, without an assisting process analysis expert. The flexibility of these systems allows easy adaptation to the continuously changing needs of a mill. Finally, we briefly discuss technological changes in papermaking which affect the diagnostics needs in the 1990's.

The psychological complexity of multivariate image quality evaluation makes it difficult to develop general image quality metrics. Quality evaluation includes several mental processes and ignoring these processes and the use of a few test images can lead to biased results. By using a qualitative/quantitative (Interpretation Based Quality, IBQ) methodology, we examined the process of pair-wise comparison in a setting, where the quality of the images printed by laser printer on different paper grades was evaluated. Test image consisted of a picture of a table covered with several objects. Three other images were also used, photographs of a woman, cityscape and countryside. In addition to the pair-wise comparisons, observers (N=10) were interviewed about the subjective quality attributes they used in making their quality decisions. An examination of the individual pair-wise comparisons revealed serious inconsistencies in observers' evaluations on the test image content, but not on other contexts. The qualitative analysis showed that this inconsistency was due to the observers' focus of attention. The lack of easily recognizable context in the test image may have contributed to this inconsistency. To obtain reliable knowledge of the effect of image context or attention on subjective image quality, a qualitative methodology is needed.

Due to the rise in performance of digital printing, image-based applications are gaining popularity. This creates needs for specifying the quality potential of printers and materials in more detail than before. Both production and end-use standpoints are relevant. This paper gives an overview of an on-going study which has the goal of determining a framework model for the visual quality potential of paper in color image printing. The approach is top-down and it is founded on the concept of a layered network model. The model and its subjective, objective and instrumental measurement layers are discussed. Some preliminary findings are presented. These are based on data from samples obtained by printing natural image contents and simple test fields on a wide range of paper grades by ink-jet in a color managed process. Color profiles were paper specific. Visual mean opinion score data by human observers could be accounted for by two or three dimensions. In the first place these are related to brightness and color brightness. Image content has a marked effect on the dimensions. This underlines the challenges in designing the test images.