Skip to main content

Real-time Monitoring and Optimization of Resource Efficiency in Integrated Processing Plants

Final Report Summary - MORE (Real-time Monitoring and Optimization of Resource Efficiency in Integrated Processing Plants)

Executive Summary:
MORE was a 36-month STREP project supported by the European Commission in the field of Nanosciences, Nanotechnologies, Materials and new Production Technologies (NMP) under the 7th Framework Programme. It aimed at identifying resource efficiency indicators (REIs) that can support operational decisions in process industries through the use of real-time data and the implementation of a dedicated online decision support system.

In recent years, indicators for the environmental impact of products and production processes have been developed and are increasingly used in the communication of companies in the process industries with the public as well as in the evaluation of alternative routes and for the decision on investments into new production facilities or revamp measures. In the MORE project, we significantly extended the available indicators and their use by:
• Defining real-time resource efficiency indicators that can be efficiently used in daily operations and are computed based upon the processing of real-time data that is available in the monitoring and control systems and from innovative analytical measurements
• Taking the step from monitoring to improving resource efficiency by providing model-based real-time decision support to plant operators and plant managers.

The project dealt with resource efficiency as a multidimensional attribute with several indicators that measure different aspects of resource efficiency.

MORE was focused on large integrated chemical and petrochemical plants with many interconnected units. The real-time resource efficiency indicators and decision support tools were developed exemplarily for this domain and tested through the implementation in four industry use cases: a refinery (Petronor, Spain), a petrochemical site (INEOS in Köln, Germany), a global supplier of specialty chemical products and nutritional ingredients based on renewable raw materials (BASF PCN, Germany) and a plant producing viscose fibers (Lenzing AG, Austria).

Very good technical achievements could be reported in the industrial cases and an impact assessment that was launched in order to analyse the economic and environmental impacts shows highly promising results:
• In the Petronor case study, the MORE partners analysed and optimised the efficiency of the distribution and the use of hydrogen as a raw material in an oil refinery. An advanced control scheme and real-time optimisation of the hydrogen system with respect to resource efficiency were implemented based on reconciled online measurements, leading to a reduced consumption of hydrogen and thus savings of cost and energy. Petronor, which is the second biggest oil refinery in Spain, already estimates the economic gain from the implementation of MORE to be between 1,000,000 to 5,000,000 €/y which corresponds to the target of 3-5% of cost savings for hydrogen production. The implementation of the MORE methodology and tools also had an impact on reducing greenhouse gas emissions for this part of the refinery by about 3.5%.
• INEOS in Köln developed an integrated REI aggregation and reporting framework for the whole site and REI calculations and visualisations for a number of production plants. The full implementation is still underway, but the work during MORE was proven already to provide the basis for economic and environmental improvements by giving better support to operators and plant managers. Regarding the environmental impact, the indirect greenhouse gas emissions (through electricity savings) could already be reduced by 16,400-32,900 t CO2eq per year.
• Lenzing, a reference factory around the world for producing man-made cellulose fibres, optimised the specific steam consumption together with the overall cycle cost of the evaporator system for the spinbaths of the viscose fibre production. Optimal control schemes and a model-based decision support system for the choice of the active evaporators and their loads have been implemented and provided significant savings of natural gas. Overall, the economic impact of all optimisation measures using MORE approaches in savings of steam consumption could lead to economic benefits of 575,000 to 825,000 € per year – more than five times the contribution of the European Commission budget to Lenzing in the MORE project. These results are also coupled to a significant effect on climate change, by reducing direct CO2 emissions from site by about 0.3%. Further investments will be made beyond the project in energy efficiency and it is planned to spread the MORE approach to other plants of the company and in others application areas, increasing the positive economic and environmental impacts.
• In the BASF PCN case several spectroscopic process analytic measurements were tested at the target plant and one of these is planned to be implemented in the plant as a substitute for manual sampling and analysis. REI calculations and visualisations were implemented and are expected to lead to efficiency improvements and economic savings.

The generalisation of the first impact results achieved in the MORE industrial cases to the process industry in Europe as a whole showed promising high potential: It is realistic to say that on EU level there is a 3-5% of saving potential in all plants through improved process operations using the MORE tools. If only 25-50% of all plants implement resource efficiency indicators and optimisation tools as developed by MORE, the economic and environmental impacts throughout Europe will be very significant.

In order to allow the management and operational staff of other industries to implement the methodology and the tools developed by MORE, a Guidebook providing step-by-step guidance has been developed and made available.

MORE has brought the computation, visualisation and use of resource efficiency indicators to a new level by evaluating them online in daily operations, visualising them for operators and managers in a transpar-ent fashion and using them in decision support and optimisation and control systems.

An Industrial Stakeholder Panel (ISP) with experts from diverse EU process industry fields confirmed the relevance of the MORE project outcomes to other industries. The final event, a Workshop on “Real-time Monitoring and Optimisation of Resource Efficiency – From Measurements to Optimal Operation” held on February 15-16, 2017 at the premises of DECHEMA in Frankfurt, Germany, provided a forum to showcase and discuss the achievements with experts from different sectors of the process industries. The advances of MORE and the usefulness of the results for all of the process industry were applauded unanimously.

MORE contributed significantly to the European 20-20-20 Targets: European policy makers introduced goals for the year 2020 in a number of different sectors. In the energy sector the 2020 goals were based on the three pillars leading European energy policy: Security of supply, competitive markets and sustainability. The 2020 energy goals are to achieve a 20% (or even 30%) reduction in CO2 emissions compared to 1990 levels, 20% of the energy, on the basis of consumption, coming from renewables and a 20% increase in energy efficiency. MORE developed efficient tools that contribute to reaching these goals.

MORE contributed also very significantly to standardisation. Triggered by the MORE consortium, Europe’s largest Association for Process Automation, NAMUR established an ad hoc standardisation group on Re-source Efficiency Indicators that will provide a NAMUR recommendation in 2017. Building on this, support from DIN DKE K931 was realised to establish a new work item proposal for an international standard in terms of an IEC Technical Report (IEC TR).

Project Context and Objectives:
The chemical industry is an important sector of the process industry: world-wide chemicals turnover was valued at 3,232 billion € in 2014 and the chemical industry worldwide accounts for more than 30% of the global industrial energy use (including feedstocks).

The chemical industry is an energy-intensive industry: energy use and transformation is a major issue in the chemical industry and its optimisation thus an important topic. In Europe, the chemical and petrochemical sector corresponds to about 19% of the industry energy consumption, according to Eurostat. In 2013, the fuel and power consumption of the EU chemical industry, including pharmaceuticals, amounted to 51.5 million tons of oil equivalent (TOE). The chemical industry transforms energy and raw materials into products required by other industrial sectors as well as by final consumers. The cost of energy and raw materials is a major factor in determining the competitiveness of the EU chemical industry on the global market.

In addition to these economic considerations, the chemical industry is also subject to a number of regula-tions, both on national and European levels, to limit greenhouse gas emissions and pollution. In order to steer the environmental actions in the chemical industry, regulatory and voluntary frameworks have been set up and implemented throughout the industry. In addition, the European Commission has set up the European 20-20-20 Targets for all industry sectors: the 2020 energy goals seek having a 20% (or even 30%) reduction in CO2 emissions compared to 1990 levels, 20% of the energy, on the basis of consumption, com-ing from renewables and a 20% increase in energy efficiency. European industry associations such as CEFIC and A.SPIRE support these goals. These policy regulations and targets are underpinned by an often strong public opinion pushing for minimised industrial emissions and efficient use of raw materials and energy.

The chemical industry has already made strong and long-lasting efforts towards resource and energy effi-ciency.

Due to intensive efforts in the chemical industry on energy efficiency, the fuel and power energy consumption per unit of production (energy intensity) has already been reduced by 55.4% until 2013 in comparison to 1990 (data on the chemical industry including pharmaceuticals). This is a considerable improvement and good performance when comparing with the EU manufacturing sector overall.

In 2014, European chemical industry’s greenhouse gas emissions totalled 131.6 million tonnes of CO2 equivalents. This equals to about 3% of the total European greenhouse gas emissions, that were 4,419.2 million tons of CO2 equivalents in the year 2014. Development of cleaner and safer technologies, putting effort on waste recycling processes and also in developing new products has not only enabled the greenhouse gas reductions but also other environmental impacts have clearly decreased over the past 25 years.

Nevertheless, the SPIRE Roadmap shows that the whole chemical industry “loses” approximately 40% of the used energy and utilises approximately 60%. It also states that physical changes are required to improve this figure; strongly energy integrated plants will present a better value. The EIA study “Potential of best practice technology to improve energy efficiency in the global chemical and petrochemical sector”, indicates the worldwide energy efficiency improvement potentials for the chemical and petrochemical sector to be approximately 18% and 21% for 2004 and 2005 respectively, if best practice technology (BPT) was applied. BPT is defined as “best practice technologies that are currently in use at industrial scale and [that] are therefore, by definition, economically viable”.

Besides high energy prices and scarcity of raw material, the regulatory environment puts additional pressure on the EU chemical industry: indeed, according to the results of the evaluation of the cumulative costs, recently undertaken by the European Commission, the total cost of legislation for the EU chemical industries amounted to €10 billion per year on average from 2004-2014. Main cost elements are regulations on industrial emissions (33%), chemicals (30%) and workers’ safety (24%). Increasing costs are expected with regards to stricter emission limit values and energy efficiency objectives as well as carbon footprint reduction.

The MORE project aimed at identifying real-time resource efficiency indicators (REIs) that can support operational decisions in process industries and at the development and implementation of visualisation, optimisation, control and decision support systems based on these REIs that lead to improved resource efficiency. Through the implementation and validation of the developed methodologies and tools in four industrial case studies, MORE demonstrated the potential of the use of real-time REIs and decision support systems in industrial operations in the chemical industry. Also, the transferability to other industrial sectors was confirmed. MORE built on existing good practices but brings them to a higher level.

Resource-efficiency and life-cycle inventory indicators are frequently used to assess the environmental impact of industrial production. Such indicators up to now have been used only in retrospect, averaging the performance over long periods of time, e.g. a year. The daily operational decisions in the plants influence these indicators, but this connection up to now was not transparent because resource efficiency was not measured over short periods of time. The goal of MORE was to monitor resource efficiency during daily operations of large production plants and to influence the operational decisions such that the environmental footprint is constantly minimised.

Towards reaching this goal, suitable indicators have been defined that provide meaningful information about the resource efficiency over short periods of time like hours or days, and new analytical measurements to provide the necessary data to be screened and tested. Based on the new indicators, decision support for the operating staff has been developed to guide the decisions towards higher resource efficiency.

The results have been implemented and validated in four industrial sites that recover a significant part of the industrial value chain of the chemical industry. Transferability to other industry sectors is promoted by the provision of a “step-by step guide” on the set up and implementation of resource efficiency indicators. Major outcomes of the project will be archived in an edited book volume “Resource Efficiency of Processing Plants -- Monitoring and Improvement” edited by Stefan Krämer (Industrial Applications Coordinator of MORE) and Sebastian Engell (Scientific Leader of MORE). The volume will be published by Wiley VCH in 2017.

Project Results:
1. Definition of real-time resource indicators (REIs)

The MORE resource efficiency indicators are based on eight principles. It is important to take these principles into account when starting to identify REIs to ensure real-time capability and that the indicators reflect the technical performance as a result of the plant operation.

• Gate-to-gate approach
As the entity of interest is a production site, a plant or a process unit, the boundary of the analysis is the limit of the respective entity, as only this can be influenced in real-time.
• Indicating technical performance independently of market fluctuations
The flows of material and energy are not to be related to real-time economic indicators; technical performance is separated from the economic performance.
• Based on material and energy flow analysis
The resource efficiency indicators are based on the physical flows and conversion of raw materials and energy to products and flows into the environment as objective characteristics of a production process.
• Resource and output specific potential for meaningful aggregation
Within the system boundaries, the indicators need to be directionally correct, i.e. improve-ments of the indicators demonstrate better process performance. All net flows of raw materials, energy, and products that cross the boundaries of the system under consideration must be determined without aggregation.
Based on a material and energy flow analysis, process specific REIs should be defined with respect to the resources and the products. The indicators can either be defined as intensities or efficiencies depending on the user preference. The definition of resource intensity is shown below. This version of the indicator simplifies the aggregation over different contributions due to having the same basis (product output). The corresponding indicator defined as efficiency is obtained by inverting the intensity indicator.

Such a resource and product specific (RPS) REI by itself does not indicate whether the process is operated well. It must be compared with a reference value obtained from historical or model data to evaluate the plant resource efficiency change:

• Considering storage effects
To realise “real-time” REI calculations, the choice of the temporal aggregation interval is cru-cial. The interval should be short enough to allow the derivation of operational decisions. Ideally a hold-up change is considered in the consumption or production figures. Long-term effects such as catalyst degradation or fouling must be defined in a suitable manner.
• Include environmental impact
The impact on the environment must be taken into account separately in order to measure the ecological performance. Emission of pollutants to air, water and soil can be used as sepa-rate indicators.
• Hierarchy of indicators – from the whole production site to a single apparatus
Production processes are interconnected. Analysing an individual apparatus may be misleading because resource utilisation can be shifted to other units by different local operational policies. Generic resource efficiency indicators must be defined on a scale where the net effect on the resource efficiency can be measured through a bottom-up aggregation.
• Extensible to life-cycle analysis
For reporting and assessment purposes, an extension to a Life Cycle Assessment should be possible using the aggregation scheme and adding a relevant weighting value to feed streams.

1.1 MORE Real-time Resource Efficiency Indicators

REIs developed in the MORE project can be used for real-time monitoring and optimisation of resource efficiency in processing plants as well as for reporting, and they are extendable to Life Cycle Assessment. Depending on the intended use of the indicators, the interpretation of the term “real-time” differs. Loosely speaking, real-time means often and timely enough for the actions that are based on the indicators. Due to the presence of disturbances and fluctuations in all production processes, resource efficiency indicators must be averaged over sensibly chosen intervals in order to avoid their values being dominated by stochastic influences. In order to properly reflect the effects of the operational policies, the averaging should generally not be longer than the periods over which the manipulated variables are kept constant.

Real-time REIs are significantly different from REIs or KPIs for historic analysis, because they allow online monitoring and rapid intervention to improve resource efficiency.
In this project a measurement, analysis, an REI or an optimisation technology is considered “real-time” if
1. The time delay and the sampling time of the entire analysis procedure – measurements and data processing – are sufficiently short compared to relevant process dynamics
2. The time resolution is similar to the typical frequency of changes in manipulated variables .
Resource efficiency indicators must be averaged over sensibly chosen intervals in order to avoid domination of their values by stochastic influences. In order to include effect of the operational policies, the averaging must not be longer than the periods over which the major manipulated variables are kept constant.

Resource efficiency indicators are classified into three categories (Figure 1):
1. Energy: This is based on an energy flow analysis (EFA). Indicators from this group measure how much energy is consumed for the production of one unit of product.
2. Material: This is based on a material flow analysis (MFA). Indicators from this group measure the amounts of raw materials consumed for the production of one unit of product.
3. Environmental: Here, the REI measures the environmental impact of the production process, e.g. by measuring greenhouse gas emission equivalents per ton of product.

Figure 1: Categories of resource efficiency indicators (see document attached)

For some indicators, the classes may overlap. The categories “Energy” and “Material” are based on energy and material flow analyses, the category “Environmental” is measuring environmental loads, such as greenhouse gas equivalents or water usage.

Resource efficiency is a multi-dimensional entity (because multiple resources are usually needed to produce a product or several products simultaneously), whereas economic efficiency can be measured by one single figure and in one single unit, money. The consumption of different resources and the environmental impact can be integrated into one figure by weighting the streams in comparable units. If these units are financial (prices or costs), the single figure comprising the weighted separate resources will fluctuate with price or cost fluctuation losing its physical meaning. If the weights are chosen on physical grounds, for example the energy that is required to produce a certain carrier of energy, such an integration can help describe resource efficiency using a single figure. Wherever possible, physical units should be preferred to make resource efficiency transparent and to reduce the influence of external factors.

In most cases, a more resource-efficient operation is also economically advantageous, but it is possible that the two objectives conflict because of the cost of measures for the improvement of the resource efficiency or external financial incentives. From the resource efficiency perspective, for example, the minimisation of all waste streams is desirable, but this can be associated with high costs, resulting in sub-optimal production from the economic point of view. In such possibly conflicting cases, REIs and economic performance indicators should be considered and reported separately, and analysed e.g. in the form of a Pareto curve or Pareto surface.

The REI approach is different for continuous and batch processes, the main difference being the non-stationary character of the indicators in the batch case.

1.2 Baseline calculation

Resource efficiency indicators should help identify whether or not the current operation is good with respect to resource efficiency. Hence their absolute value is often not as important as the comparison to a reference point which is called baseline in the following text.

We define Baseline as the performance of the plant averaged over a specified period of time. Typically, the average of production data over a representative period in stationary operation is used. The value of the baseline usually depends on the mix of products in the period considered, the raw materials used, the operation regime (e.g. load levels) and external influences, e.g. outside temperature. Therefore, the baselines should, if possible, be differentiated with respect to the main factors that are externally set and cannot be influenced during operation.

The Best Demonstrated Practice (BDP) is an important baseline. It is defined as the best observed operation based on historic data. In case of batch production, the BDP is also called a golden batch. As mentioned above, dependencies of the BDP on the product mix, load levels, raw materials and other influences are often observed.
A term Best Achievable Practise (BAP) is used if the baseline is an optimum that is computed from an analysis of the limits of the process performance. In such a theoretical analysis, also the influence of external factors can be considered.

The external factors that influence the baseline, the BDP and the BAP, can be separated into influenceable and non-influenceable factors. If the baseline is not adapted dependent on the key non-influenceable fac-tors, comparing different operation points becomes very difficult.

The effect of the non-influenceable factors can be visualised by a time-varying baseline or BDP corresponding to the external conditions. Different people in a hierarchy have different decision authorities and influence on a plant and thus their baselines may reflect different influenceable and non-influenceable factors. Table 1 illustrates this idea. As an example, the operators cannot normally improve plant operations by buying better or different feed stock, but the site manager might be able to make such decisions.

Table 1: Influenceable and non-influenceable factors for different management levels (see document attached)

A simplified representation of BDP baselines for the operators and for plant management is shown in Figure 2 (see document attached).

Figure 2: Different BDP baselines for different groups (see document attached).

1.3 Real-time Resource Efficiency Indicators for batch processes

Batch processes have different characteristics, due to the time-varying nature of the processing steps and the integrity of the batches. This requires modifications of the real-time resource efficiency indicators compared to the ones introduced for continuous production processes. The main difference is that the appropriate level of granularity for batch plants is an individual batch or a number of batches. In order to determine the resource efficiency of the production of a single batch, the resource consumption must be measured with such a resolution that the resources can be attributed to the individual batch.

The efficiency of batch processes is affected on three levels:
• Batch level: All contributions to the emissions and the consumption of materials or energy that is recorded at the equipment with high granularity in time is directly attributed to the batch currently processed in the equipment
• Logistic level: All contributions that cannot be directly attributed to a single batch are recorded within overall resource efficiency indicators (ORE). Effects on this level can be influenced by the chosen production plan or slow degradation processes of the equipment or catalyst. OREs need to be evaluated over longer time horizons.
• Production phase level: In order to supply additional information that might be useful in a root-cause analysis, unit specific indicators can be defined for key production steps, e.g. the reaction phase or the purification step. Such indicators are useful to find the reason behind inefficient pro-duction states that are identified by indicators on the batch level, but might not be aggregated in the batch efficiency or plant efficiency measures due to double accounting.

2. Visualisation

Visualisation techniques are used to present data to the user in a somewhat abstracted representation that is intended to ease the interpretation process by giving meaning to data and supporting an efficient perception. The first step towards a good representation of the resource efficiency of a chemical production plant is the definition and selection of key indicators that accurately represent the true plant performance. This was previously accomplished in the form of resource efficiency indicators (REIs) within the deliverables D1.2 for continuously operated processes and D1.3 for batch processing plants. The next step is to find the visualisation techniques that represent the REIs best on a human-machine-interface.

In general, a well-designed dashboard conveys information that is :
• Well organised
The representation of complex systems requires multiple REI and supporting information, thus multiple plots and graphs are needed. Grouping in rows, columns fields increase readability and reduce the time required to find the needed information. Texts written in the languages of the participating project partners are read from left to right and up to down. Taking advantage of this habit the most important information should be displayed in the top left area and the least important on the lower right (cf. Figure 3).

Figure 3: Structure of well-organized dashboard configuration (see document attached).

• Condensed, primarily in the form of summaries and exceptions
Information in the form of hundreds of data points is not helpful in the interpretation process. The defined REIs are in fact summaries over meaningful time scales and subsystems making it possible to see the whole picture. Data plots show plant behavior in the bounds of normal operations should be displayed unobtrusive (light colors with low intensity), in order not to draw the attention away from information that is more important. More intense colors and forms of representations should dynamically occur in the case of exceptions (a situation that requires the operator evaluation or input).
• Specific to and customised for the audience and objectives
The kind of information presented to the recipient should be based on what is needed to execute the task at hand. Insight in what this information is can efficiently be gained by observing the operational procedure in the current set-up and questioning why and on what basis a decision was made.
• Displayed using concise and often small media that communicate the data and its massages in the clearest and most direct way possible
The content and density of information in a visualisation element should be limited to a minimum which still serves the intended purpose. This includes aspects like background coloring, background pictures and unnecessary grids and borders. These rules were introduced as the concept of “data-ink ratio” by Edward R. Tufte , stipulating the reduction of non-data ink.
Based on these requirements to dashboard representations S. Few identified 13 common violations of these rules that should be avoided in the design of effective visualisation solutions.
• Exceeding the boundaries of a single screen
• Supplying inadequate context for the data
• Displaying excessive detail or precision
• Choosing a deficient measure
• Choosing inappropriate display media
• Introducing meaningless variety
• Using poorly designed display media
• Encoding quantitative data inaccurately
• Arranging the data poorly
• Highlighting important data ineffectively or not at all
• Cluttering the display with useless decoration
• Misusing or overusing color
• Designing an unattractive visual display

Below one visualisation element is discussed that is useful in the visualisation of REIs.

2.1 Example for a smart visualization element

Figure 4 (see document attached) shows a bullet chart that conveys condensed information about the current value, the immediate history, and the extrapolated prediction of an REI . The REI should be defined with respect to a theoretical optimum or best achieved value (100%). The white interval of the scale indicates the interval above the target value, the grey interval below the target value respectively. The lower bound of the scale should be chosen as 0% to achieve a consistent and comprehensible visualisation. In case the desired domain of operation is very close to the optimum for all indicators, it is useful to choose a higher value as lower bound for the scale. Triangles are used to mark the current value and are complemented with the numerical value on the opposite and an arrow designating the direction of movement based on the immediate past. The colored rectangle is the variability bar and shows the range of values that was exhibited during a time period indicated on the lower right. In case an indicator leaves the white interval, an exception occurs that is emphasized by a change in color of the variance interval/numerical value and the appearance of a caution sign above the scale.

The history of the indicator is stored implicitly in the size of the variability bar, the current position relative to the variability bar and the arrow indicating the direction of movement. If the plant operates stable within the desired efficiency range the variability bar is small and lies entirely in the white area (reactant efficiency). An upset plant manifests in large variability bars that may reach into the sub-target range (Material efficiency). A triangle position at the border of the variability bar in combination with an arrow pointing further away from the variability bar indicates a transient trend away from the former average (waste efficiency) which can be an early indicator for the operator to intervene and take corrective measures. Finally, the color change and appearing warning signal drag the user’s attention to the state. Bullet charts do not necessarily have to be oriented from bottom to top, but could also be displayed top to bottom (in a minimisation task) or with an orientation from left to right, depending on the available dashboard space and the context they are used in.

2.2 Selection of visualization elements for efficient concepts

In the previous section one data visualisation element was presented as an example for a larger number of visualisation elements that were developed during the MORE project. A comprehensive overview over the complete set of elements and their suitability for different requirements is given in Table 2. If the considered element fully meets one of the requirements listed on the left, then this is indicated by a “+” sign in the corresponding field of the matrix. A “•” sign is used if the criterion is partially met and the field is left blank if the form of presentation is not suitable for the requirement. If the selection criteria for a planned visualisation task are defined, Table 2 can help to select appropriate methods to create the most efficient dashboard solution possible.
In most cases, it is not possible to display all aspects equally excellent in just one diagram, thus it is beneficial to use methods alongside each other that highlight different aspects, i.e. a plant structure diagram with bar indicators for the total efficiency of the section can be used to show the overall state, along with stacked bar charts that further break sown the contributing factors of the overall efficiency. With a smart selection of visualisation techniques that built on one another and highlighting important data, exceptionally efficient HMI can be created.

Table 2 (see document attached).

A prototypical MORE REI visualisation dashboard that was designed for the case of a sugar plant is shown in figure 5 (see document attached).

3. Analytics and data reconciliation


Activities in the field on analytics aimed at meeting the needs of the computation of REIs in terms of real-time measurements:
1. Which measurements are currently missing when REIs are being calculated?
2. What are the accuracy requirements?
3. Are suitable analytical technologies available, and how can they be trained most efficiently?
4. Can these techniques be implemented under industrial constraints?

For two of the Case Studies, tasks for PAT technology were specified from the requirements to compute resource efficiency indicator. These included:
• Composition analysis of a complex hydrocarbon mixture with > 50 components, enabling more dy-namic feed management to the plants (INEOS Case)
• Residual substrate content monitoring in a batch reaction involving solids, enabling faster end point detection (BASF Case)
• Substrate removal monitoring in a separation step, enabling more gentle process conditions and relieving thermal stress from the product (BASF Case)
• Colour measurement as a quality monitoring in pre-final processing step, enabling a reduction of chemicals input in the subsequent treatment step (BASF Case)


The hydrocarbon analysis was achieved with process Raman spectroscopy, a technique beyond the state-of-the-art online near-infrared (NIR) or offline GC-MS techniques. It resulted that chemometric method training was most efficient and flexible with measurements from a lab-scale process simulator.

With respect to the complex composition of petrochemical hydrocarbon mixtures, an approach was devel-oped to come to a selection of training samples covering the calibration range as uniformly as possible. With the ranking approach applied, prediction methods for the 15 features most relevant to resource efficiency were trained on 80 well-distributed material samples only.

A Raman process analyser was temporarily installed in the field to demonstrate the online analysis system in a realistic industrial environment (see figure 6). The achieved accuracies were in the range of max. 0.5% for the major components and substance classes, and down to 0.1% for the individuals (cf. Figure 7 for the 5 component classes P, I, O, N, and A).

Figure 6 (see document attached)

Figure 7 (see document attached)


From the three analysis tasks defined from the REI requirements, for two applications successful solutions were provided by S-PACT.

The monitoring of the residual substrate in a batch reaction turned out to be challenging when solids were present. For products being produced from a fully homogeneous reaction, a Raman spectroscopic monitor-ing turned out to be feasible, but did not provide sufficient benefit to end point detection due to a lack of accuracy.

Online monitoring of substrate removal was demonstrated successfully, so that thermal stress to the prod-uct and product degradation could be reduced. However, the benefit was too small to justify the implementation of the method permanently.

The application of an online spectroscopic colour measurement significantly improved the monitoring of the post-processing step for the products, and led to a substantial reduction of the chemicals needed for the product treatment. The main benefits evolved from the better time resolution, the improved resolution on the colour scale, and the objectivity of the colour measurement that was no longer affected by colour perception of the operator. Figure 8 shows results which were obtained with the UV-VIS measurement.

Figure 8 (see document attached)

Data Reconciliation

Formulations of data reconciliation problems adapted to the requirements of the MORE industrial case studies were developed for the Petronor and Lenzing cases, and initial work was done for the INEOS case.

The Fair Function was selected as robust estimator for the final implementations, leading to a formulation of the data reconciliation problem as indicated in the formula (see document attached).

DR algorithms were integrated into the prototypical tools for optimisation and decision support which have been developed for the case studies.

A general methodology was proposed covering all aspects in the data reconciliation problem, from the model-identification stage to the real-time implementation. The proposed procedure is as follows:
1. Perform a first step of data treatment to exclude outliers (faulty sensors, out-of-range values, etc.)
2.Average data according a suitable time window or run a steady-state detector if necessary to ex-clude transient evolutions.
3.Perform data reconciliation with the Fair function as objective to get a set of reliable process values and model parameters.
4. Identify experimental patterns and relations among variables by inspection, clustering or any avail-able identification method. Formulate regression constraints and incorporate them into the original model.
5. Validate the obtained grey-box model with new data reconciliation.
6. Embed Steps 1 and 2 together with the robust data reconciliation (Fair function constrained to the above identified grey-box model) into a software module to be executed in real time.

As a result of the development of the above procedure, a software module for data reconciliation has been developed to interact with the decision support (optimisation routines and operators) and with the DCS system. This provides reliable information to the tools for REI visualisation and decision support (see Figure 9).

Figure 9: Interconnections between data reconciliation, simulation and optimization with the plant DCS (see document attached)

4. Decision support

Decision support tools were developed within MORE on different levels of sophistication:
• Simulation-based what-if analysis
• Real-time stationary optimisation of resource efficiency
• Dynamic optimisation of resource efficiency
• Decision support for equipment allocation using mixed-integer optimisation.

4.1 Simulation-based what-if analysis

A prototypical tool for what-if analysis has been developed, which provides scenario based decision support in terms of resource efficiency. The tool assists operators and plant managers to define and manage scenarios, to (semi-)automatically evaluate them and to visualise simulation results.

In order to store simulation results for future reference, it is possible to save and load the analysed what-if scenario including all numerical values of inputs, outcomes and indicators plus a description using a XML scheme designed within the task.

The methodology has been applied to a cooling tower case-study supplied by INEOS for three scenarios to show potential energy savings when the ambient conditions (air temperature, pressure, and humidity) are considered. A weather forecast enables the operators to act even anticipatory and get around possible limitations due to environmental conditions.

What-if analysis tools were also developed for the Lenzing and Petronor case studies.

The models of both case studies were simulated in the environment EcosimPro, adding an OPC server layer to the executable code later on. A SCADA system was configured and connected as an OPC client to the simulation, which was able in this way to interchange values with the plant SCADA system. The HMI allows to specify values of parameters and boundary variables as well as reading them directly from another source (e.g. real-time control system or data reconciliation estimations) in the simulated models. The SCADA collects and displays the simulation results in different formats. A screen for an evaporation plant at Lenzing is shown attached in Figure 10).

Figure 10: User interface for what-if analysis for evaporators (see document attached)

4.2 Real-time stationary optimization of resource efficiency

Real-time optimisation of resource efficiency means a model-based optimisation of the operational parameters using a rigorous model of the plant with respect to resource-efficient operation and respecting technical and operational constraints, e.g. limits of process variables, quality parameters, and production targets.

In the Petronor case study, the system considered covers eighteen hydrogen producing and consuming plants and the related distribution network. The operational framework is delimited by the refinery planning, that establish ranges for the hydrocarbon loads to the treatment plants as well as for the desulphurisation levels. The following targets for the optimal network operation were considered, which as weighted in the cost function according to their relative importance:
• Maximise the production targets of hydrocarbon feeds to the plants within the ranges established by the refinery planning system.
• Balance the hydrogen that is produced and the hydrogen that is consumed so that the hydrogen losses to fuel gas are minimised.

Being the main decision variables, the generation of fresh hydrogen in the producer units and the distribu-tion of the fresh and recycled hydrogen, including the use of purification membranes. The formulation of the optimisation incorporates additional constraints oriented to keep the operation of the control rooms as undisturbed as possible. Because of that, it assumes as fixed quantities many specific values related to the current operation of the units, such as specific hydrogen consumption in the reactors or other properties in separation units. In the same way, the state of functioning or stopping the plants and the current structure of the network are respected, assuming that they are mainly imposed either by maintenance or global pro-duction planning reasons. The resulting problem is a large nonlinear program. The final implementation of the optimisation problems was made in the GAMS environment for the Petronor case, using IPOPT as the NLP algorithm.

Real-time multi-criterial optimisation

A tool for multi-criterial optimisation was developed and tested on the INEOS case study of a butadiene plant that is coupled with a cooling tower with multiple cells. The multi-criterial optimisation implemented in the tool is performed by an evolutionary algorithm (EA), which is a variant of the NSGA-II algorithm and implemented in MATLAB’s Global Optimisation Toolbox. An evolutionary algorithm was chosen because of its applicability to a wide range of problems and a good performance in terms of finding near-optimal solutions.

For the particular application, a three-dimensional optimisation (steam consumption, electricity consump-tion, and solvent loss) was performed in order to yield the set of Pareto optimal solutions for a desired production capacity. Based on the results that are displayed in different interactive visualisation elements, as exemplarily shown in Figure 11, the user is able to make an informed decision about the preferred operation conditions.

Figure 11: Visualisations for three optimisation criteria (see document attached)

Thus, it was shown that multi-objective real-time optimisation of the combined plant complex is possible. The weather forecast can be included to obtain the optimal operation conditions at any time.

4.3 Dynamic optimization of resource efficiency

The real-time optimisation (RTO) approach is based on static models and provides the best stationary mode of operation according to the conditions considered in the formulation of the optimisation problem. The optimal steady state is then implemented via the plant control layer which also rejects disturbances that act on the plant and reduces the influence of varying parameters as e.g. feed properties, weather conditions, fouling etc. The interactions between the decision variables of the control layer and of the RTO layer and the need to respect process constraints at all times and to continuously react to disturbances requires a close cooperation between the RTO and control layers. A strategy for integrating RTO and control has been studied and applied in the Petronor case (see figure 12). It is based on the analysis of the solutions given by the RTO system which led to the recognition of patterns that were then implemented in the dynamic control layer taking advantage of the architecture of existing model predictive control (MPC) systems.

Figure 12: Optimization approach using DMC for the Petronor application case (see document attached)

In particular, in the Petronor case, a commercial dynamic matrix controller (DMC) is available that combines an LP optimiser with a dynamic DMC controller that receives the targets at the end of the prediction horizon from the LP layer, operating with a sampling period of one minute. The cost function of the LP optimiser implements the optimal policies that were extracted from the RTO, which are applied in a dynamic context considering the changing disturbances, constraints, etc. These policies are translated into local targets, such as to minimise recycle purities, losses to fuel gas from the low purity header, etc., instead of specifying values of set points.
In the Lenzing case, such optimal patterns could be directly implemented in the DCS control structure, as they correspond to maximum or minimum values of the decision variables, such as juice temperature or vapour temperature to the cooling tower.

A prototypical dynamic optimisation tool was developed for the butadiene plant of the INEOS case study. The multicriterial steady state optimisation provides the starting point of the dynamic optimisation. On the upper right side of the interface, shown in Figure 13, the user can select an operating point from the set of Pareto optimal operating points (green dot) to which the plant should be moved from the current operation (blue diamond).

Figure 13: User interface of steady-state optimisation tool for the selection of the desired operational point (see document attached)

The dynamic optimisation problem is formulated as a typical MPC optimisation that considers three per-formance measures in the objective function to be minimised:
• The difference between the reference value for the output to predicted states
• The penalty on the input changes
• The difference of the predicted inputs to the optimal inputs from the steady state optimisation

Furthermore, the user may choose the number of inputs that are optimised and the length of the prediction horizon.

In the case of a semi-automated plant operation the optimisation algorithm is free to change every input at every time step. In the case of a manual plant operation, the operator may be unable to change every input at every time step if the number of inputs is high. Therefore, a second optimisation solves a mixed-integer dynamic optimisation problem. This additional layer determines the best combination of a user specified number of inputs to get as close to the desired steady-state as possible. MATLAB’s genetic algorithm is used to optimise the set of utilised inputs which applies heuristic rules to improve a calculated solution step by step.

4.4 Decision support for equipment allocation using mixed-integer optimization

When for the Lenzing evaporation plant the scope of the problem is enlarged from one evaporator to the whole set of evaporators, more than 25 in total, that are treating several different flows, the RTO problem becomes a scheduling problem because the evaporators can be assigned to different spin-bath cycles and the assigned load can be varied, or an evaporator can remain idle. In addition, the cleaning of every evaporator should be performed at the right time so that the total evaporation capacity for every product is maintained and the total steam and cleaning costs are minimised.

The decision support that was developed for discrete decisions in the Lenzing case study relies on a semi-automatic modelling approach that is based on historical data to reduce the modelling effort. A set of evaporator models describes the evaporation network at Lenzing and continuous optimisation problems for the evaporators are solved with respect to energy-optimal operation. The scheduling of the evaporators, specifically the allocation of evaporators to the spinbath cycles and the load allocation to the evaporators, introduces integer decision variables, resulting in a mixed integer non-linear programming (MINLP) problem.

The optimisation problem at hand is demanding due to the size of the combinatorial problem that is posed by the evaporation network and the constantly changing production constraints as a result of the varying ambient temperature, air humidity, cleaning actions, maintenance restrictions, and desired evaporation capacity. For optimisation based decision support in real-time, the computational time constraint is tight, because the results need to be available with an acceptable delay, to ensure operator acceptance and to avoid delays in production. Thus, an as simple as possible model for the efficiency of each evaporator is called for that can easily be updated if components of the evaporator are replaced or altered.

The fouling processes that affect the evaporators require an online adaptation of the models to the actual state of the evaporators. In order to make the approach economically viable, it is necessary to keep the modelling and maintenance effort as small as possible. The possibility to perform experiments is limited, since an interruption or limitation of the production capacity is costly. As a result, a data-based modelling approach that is based on available historical data was selected.

Figure 14: Structure of the modelling and optimization approach for operator decision support in the Lenzing case study (see document attached)

Figure 14 (see document attached) shows the integrated modelling and optimisation approach using historical and current production data from the process control system and from the data historian. The semi-automatic modelling procedure for every evaporator is shown with a dark grey background. It is performed once offline to compute the model parameters that are time-invariant and reflect the physical set-up of the evaporator under consideration. The available data undergoes a data treatment step that removes inconsistent data sets. The remaining viable data is screened for step changes in the evaporation capacity and extracts the stationary production information before and after the detected step. A least squares optimisation is performed to find the best parameter fit for simple regression models of the specific steam consumption and evaporation capacity for each evaporator.

The fouling state is a dynamically changing influence and needs to be updated before each optimisation run. The light grey area in Figure 14 highlights the periodic optimisation procedure of the operational conditions and evaporator allocations. If one execution of the optimisation routine is triggered, an algorithm is executed that
• compares the model prediction, based on the current evaporator inputs, with the measured evaporation output and steam consumption to derive the actual fouling state,
• calculates the desired evaporation capacity per cycle from the production demand and translates these into constraints for the optimisation problem, and
• updates the information on the evaporator network as a result of maintenance and cleaning efforts.

The collected information about the model parameters, the fouling state and the current constraints is then used to set-up a mixed integer non-linear optimisation problem (MINLP) that is solved to (local) optimality. The optimisation results are displayed to the operators that will subsequently adjust the network operation accordingly. This decision support solution has been integrated into the IT-systems at Lenzing in the form of a prototypical tool.

5. Implementation platform

The computation of REIs from measurements is not very complicated, provided that the necessary meas-urements are available and are sufficiently accurate. The computations can be implemented in state-of-the art PLC (Programable Logic Control Systems), DCS (Distributed Control Systems), SCADA (Supervisory Control and Data Acquisition Systems), MES (Manufacturing Execution Systems) or PIMS (Plant Information Management Systems. However, the one-by-one implementation of each individual calculation for each unit considered is time-consuming and the maintenance of these computations when changes are made to the instrumentation or to the IT systems is problematic. The new IT approach of MORE is to use formal information models for real time REI applications. The goal of information modelling is to provide a process-oriented context to the data sources in order to combine real-time data with additional information based on the structural information. An implementation of REI applications based upon an information model of the plant or site can reduce the effort considerably in the long run.

The goal of such an information model is to provide a process-oriented context to the data sources in order to combine the data with additional information based on the structural information. The knowledge of the material and energy flow network of a site or plant is of special importance and relevance. Besides the amount and the direction of flows including internal recycles, information about flow categories, composi-tions of flows and possibly additional characteristics are required. In addition, the following issues have to be linked to the flow and their properties data: generic calculation algorithms, characteristic constraints and also material or energy type specific master data (e.g. standard enthalpies). The knowledge that a specific measurement represents a flow of the product x, which flows between the plant unit 1 and the plant unit 2 and has product characteristics which depend on a specific property of the flow downstream is missing in today’s tools. The idea of the MORE IT approach was to use such information for automated balancing, REI calculation and aggregation and also for data reconciliation.

In order to use structural information online for REI calculations as well as for retrospective material and energy efficiency analysis, the streams in a plant should be described in a formal manner by a resource flow information model (s. Figure 15). Based on this information model, a company can describe the energy and material flows of a site in an intuitive but nonetheless formal manner so that all the information which is needed for real-time energy or material performance calculations is merged and related to a structured holistic knowledge base. This knowledge base can also be used to calculate missing measurement data or to provide online plausibility checks based upon material or energy balance equations.

Figure 15: Describing the resource flows of a site in a formal manner using a graphical editor of the deployment platform (see document attached).

Such a modelling framework was developed in MORE by the partner LeiKon as part of their production assistant tool IntexcSuite. The platform acts as a central data and calculation hub, which attaches to the data semantic information and the necessary context. REIs can then be aggregated automatically using aggregation rules based on the plant hierarchy that is described by the resource flow model. A tool to define and implement such an information model can be used as an add-on to existing Plant Information Management Systems (PIMS) or Manufacturing Execution Solutions (MES).

Building a resource flow model

In the first step a company specific Type Model must be designed. The user must define what kinds of plants and plant units, what types of resource flows and what types of substances exist at the site. In addition, the user can decide what kind of properties and calculation methods should be designed on a type level. All properties and calculation methods which are specified on a type level are valid for all instances which will be derived from a specific type element. Furthermore, an organisational hierarchy should be specified in order to define balance volumes that correspond to the organisational structure and suitable REI aggregation functions. In the second step the Type Model is used to define a site or plant specific instance of a generic resource flow model. The resource flow model is typically structured hierarchically and provides an overview of the network of interconnected resource flows and their properties. In the third step, specific properties can be linked to the units and flows. For example, a property which represents a measurement point can be linked to an external data point of a connected Plant Information Management System (PIMS) or Distributed Control System (DCS). At this engineering step, the resource flow model will be linked to real time data (see figure 16).

Figure 16: Linking real time data to elements of the resource flow model (see document attached)

Each unit or flow property can be linked to a data point of any IT System. The deployment platform offers an easy real-time access using embedded interfaces of standard communication technologies like OPC, ODBC or Web Services and also native interfaces of common tools like OSI PI from OsiSoft, IP.21 from AspenTech or SAP.

User Interfaces

In a final step, the visualisation of the calculated REIs must be designed and engineered in an application-specific user interface (s. figure 17). Typical user interfaces to monitor or analyse REIs can be dashboards or decision support tools that include interactive interface elements for the users. The dashboards can be used to steer the production plants in a resource efficient way and to report the energy and resource consumption e.g. for ISO 50001:2011, ISO 50006:2016 and ISO 14001:2015 management activities.

Figure 17: Typical Dashboards for REI monitoring and decision support (see document attached).


The key innovative approach of the REI implementation platform of MORE is to use knowledge of the proc-ess-oriented context of existing measurements to obtain REI application in a comprehensive and holistic way. The effort to develop and maintain comprehensive REI applications can be reduced considerably in the long run. Based on resource flow structure and generic calculation methods application can grow step-by-step, transparent REI calculations can be obtained on a semantic level and MCA (Material flow cost accounting) and LCI (Life Cycle Indicators) can be calculated automatically by indicator propagation. Based on a resource flow structure and given generic constraints REI aggregations can be done automatically, missing measurements can be calculated online and data reconciliation can be used in real-time to improve REI results.

6. Industrial applications results


The Lenzing site (Figure 18 attached), located in Lenzing, Austria, is a reference factory around the world for producing man-made cellulose fibres. These advanced fibres are broadly used from home textiles to medical and technical applications..

Figure 18: Lenzing site in Upper Austria (see document attached)

The production of high quality viscose fibres from pulp is a mechanical and chemical process that proceeds in multiple steps illustrated in Figure 19. A key process step in the fibre production is the spinning step, where the dissolved cellulose xanthogenate is metered through spinnerets into a sulfuric acid spinbath to regenerate viscose fibres. During the spinning, water accumulates and dilutes the H2SO4-containing spinbath. To maintain a certain spinbath concentration as well as to extract the co-product sodium sulfate, the spinbath has to undergo a sophisticated recovery cycle. An energy intensive process within the recovery cycle is the evaporation of the accumulated water. The evaporation process is performed in several multi stage evaporators, using live steam as main energy/heat source. Any reduction of the evaporator steam consumption results in significant energy savings.

Figure 19: Process steps of the viscose fibres production (see document attached)

The goal of Lenzing within the MORE project was to use meaningful REIs and DSSs to reduce the energy consumption of the spinbath evaporation process by 1%. In order to achieve such energy savings Lenzing divided the project into three different tasks.

Task 1: Improving the energy consumption of a single evaporator

The three main control variables for the operation of the evaporators at Lenzing are the circle flow rate, the feed temperature after the live steam heat exchangers and the cooling temperature of the attached condenser. Before the MORE project, the operators had to choose suitable combinations of the manipulated variables of the evaporators in order to achieve the requested evaporation capacity. During the project the use of the specific steam consumption (REI ID001) and its visualisation (Figure 20) resulted in the development of an optimal operating pattern.

Figure 20: Energy optimal operation pattern for different evaporation capacities (see document attached)

The REI visualisation clearly demonstrated that for energy optimal operation the cycle flowrate had to be as low as possible. A more efficient performance control uses the feed temperature to adjust the evaporator capacity. If and only if the feed temperature is at its upper limit, the cycle flowrate is increased to meet the evaporation capacity requirement. The more efficient operating pattern has been fully implemented as a new evaporator performance control in the DCS. Furthermore, a new temperature control for the evaporator cooling towers was also implemented in the DCS. The new control allows lower cooling temperatures, which leads to a lower steam consumption of the evaporators, while still preventing the freezing of the cooling tower.

The implementation of the two new controls in the DCS already improved the overall energy efficiency of the spinbath evaporation section by 3.0 %. This figure implies overall energy savings in the form of natural gas savings of more than 1.25 Mio Nm³ per year and a production cost improvement of more than 375.000 € per year.

Task 2: Optimising the evaporator load allocation

For the evaporation process Lenzing uses a large evaporator network that consists of several spinbath cycles and more than 25 different evaporators. Every evaporator varies in capacity, energy efficiency and connectivity. During daily operations, the operators must decide where (cycle) and how (capacity) to use every evaporator in order to achieve the required evaporation capacity per spinbath cycle. In the MORE project a decision support system (DSS) was developed to support the operators to achieve the most energy efficient evaporator load allocation.

The decision support system (DSS) for the optimisation of the evaporator load allocation uses an integrated modelling and optimisation approach that was developed in cooperation with TU Dortmund using historical and current production data from the process control system and from the data historian as well as a MATLAB optimisation tool to solve the optimisation problem. For the indication and visualisation of the optimisation results a PI-dashboard was also developed. The PI-dashboard is depicted in Figure 21.

Figure 21: Prototype of the PI-interface for the optimization of the evaporator load allocation (see document attached)

The PI-dashboard displays to the operators the actual state of the evaporator network and the proposals (results) for a more efficient load allocation according to the optimisation tool. Furthermore, small picto-grams show the necessary direction of the change in evaporation capacity. The dashboard can also be used to manually start the optimisation and it displays the predicted network-wide savings potential in steam and € saved per hour to create an incentive for the operators to apply the predicted set points to the plant.

Task 3: Optimising the evaporator cleaning cycle

The spinbath is an acidic solution that contains unknown number of impurities. Some of these impurities settle within the heat exchangers that are preheating the spinbath cycle stream before it is fed to the first evaporation stage. During operation, a deterioration of the efficiency can be observed that is attributed to this fouling process. The fouling layer is removed periodically by different cleaning procedures where the equipment is flushed with a cleaning solution. The cleaning scheduling is performed by a supervisor based on his experience and availability. In the MORE project a decision support system was developed to achieve a more efficient cleaning scheduling in terms of cleaning cycle time and cleaning sequences.

During the project a reference run was successfully implemented in the DCS to identify and analyse the fouling behaviour and the effects of the different cleaning types on the evaporators. Based on the collected data a visualisation of the normalised average cost per time (internal Lenzing KPI) was possible and has been used to find a new optimum for the cleaning cycle time. In cooperation with UVA a first prototype optimisation Excel-tool based on the reference run database has been developed. The interface of the tool is formed by several sheets (one for each evaporator) where there is a fixed set of values that the user is allowed to set: duration of cleaning tasks, costs of resources, energy prices plus all fouling and cleaning parameters (see Figure 22).

The tool can be used as a DSS by the cleaning supervisors, defining for the supervisors the optimal cleaning cycle time for every cooling tower evaporator as well as the preferred cleaning type for an improvement of the normalised average cost per time that will result in an overall lower energy consumption of the evaporator section.

Figure 22: Interface of the Excel-based tool for the improved scheduling of the evaporator cleaning (see document attached)


For Lenzing, the MORE project was a large success. Lenzing achieved important improvements in their spinbath evaporation process that lead to large energy and greenhouse gas emissions savings. Although Lenzing finished not all cases yet, they still accomplished by far their project goal and can already reap the expected benefits. These savings are in detail:

• 2.5 % savings in evaporator steam consumption, equivalent to around 1.0 Mio Nm³ natural gas per year through the implementation of the new evaporator control scheme in the DCS
• 0.5 % savings in evaporator steam consumption, equivalent to around 0.25 Mio Nm³ naturals gas per year through the new cooling tower temperature control strategy.
Furthermore, the MORE project laid the foundation for future improvements through the development of:
• A prototypical decision support system for the evaporator selection and load allocation supporting the operators to achieve the most energy efficient load allocation
• A prototypical decision support system to define the optimal scheduling of the evaporator cleanings supporting the supervisors to perform the most energy efficient cleaning cycles.

REIs were implemented to support the findings. The implementation and visualisation of these carefully chosen REIs for decision support and optimisation help Lenzing to operate more efficiently:
• REI measurements and evaluation of the evaporator fouling behaviour and the effects of different cleaning procedures.

Lenzing is convinced that projects such as MORE achieve results that impact both the company and the European chemical industry. The parts of the projects that could not be finished will be tackled in 2017 partly funded by the new H2020 SPIRE project CoPro.


Efficiency in the use of hydrogen as raw material in the oil refinery of Petronor (Muskiz, Spain) is used as a case study. In recent years, H2 requirements as reactant in many reactions have experienced a steady in-crease. Hydrogen is generated in producer plants and distributed through a complex network to the con-sumer plants. In the Petronor refinery the network includes four producers and fourteen consumers; a schematic of the plants and distribution network can be seen in Figure 23. Decisions regarding H2 manage-ment are complex as many plants and operating constraints are involved in the network operation with a high degree of interaction. In addition, feedstock usually changes every two or three days, as a consequence, scenarios regarding H2 consumption in individual consumer plants can experience frequent significant changes, therefore, REIs intended for real-time decision-making purposes are interesting to be monitored.

Figure 23: Scheme of the refinery H2 network (see document attached)

For technical reasons, the reactors in the plants have to operate with excess hydrogen, which can be recy-cled and reused according to the hydrogen purity requirements of the different consumer plants. In this context, efficiency in the operation means adjusting hydrogen production and consumption as well as distributing the hydrogen through the network and reusing the excess in such a way that hydrogen losses to the fuel-gas network of the refinery are minimised. Total cost will benefit of an efficient hydrogen operation as the production of high purity H2 is expensive. In addition, an efficient use of hydrogen use allows increasing the hydrocarbon load to the treatment plants generating further benefits.

In order to compute properly the Resource Efficiency Indicators and to take decisions that improve the process performance, reliable process information is needed. Nevertheless, there are many unmeasured important variables or unreliable ones. To solve this problem, a model of the hydrogen network has been developed and used in a data reconciliation system, to estimate consistent values of all network variables. The data reconciliation step takes advantage of the redundant instrumentation and provides sensible estimations of previously unmeasured variables like hydrogen consumption in the reactors or molecular weights of the streams. Additionally, an indication of potentially faulty instrumentation is obtained.

Once adequate process information is available, the model can be used in a second step to compute the best way of generating and distributing the hydrogen and increasing the hydrocarbon load to the consumer plants. This is performed solving an optimisation problem that considers the available degrees of freedom, the process constraints and the specifications of the refinery planning system and provides indications about the hydrogen that must be generated, how it should be distributed through the network and used in the plants and the best hydrocarbon loads to each plant. The data reconciliation and network optimisation software is linked to the PI information system of the refinery, as can be seen in the left-hand side of figure 24. The system runs every 2 hours reading process values and sending back results of REIs, reconciled data and optimisation recommendations.

Figure 24: Architecture of the network optimization system (see document attached)

For real-time implementation of the optimisation, optimal policies proposed were analysed and a predictive controller, a DMC one, executing them was developed and deployed covering two hydrogen producers and the four main hydrogen consumer plants. The DMC runs every minute and takes into account the process constraints and the targets computed by its LP layer as can be seen on the right-hand side of figure 24.


The different modules implemented have improved the operation of the hydrogen network in several as-pects, being the most important ones:
• Better information about the process state
• Savings in the use of hydrogen
• Increments in the hydrocarbon processed.

Better process information is the result of several factors. Analysing systematically the process measure-ments and performing data reconciliation at regular time intervals of two hours provide:
• An indication of possible faulty instruments
• Reliable balances of hydrogen
• Values for unmeasured quantities (purities, molecular weights, hydrogen consumption...) not available previously
• Data for computing REIs that allow better monitoring of the operation of the network.

The integration of the results of data reconciliation, REIs and optimal operation into the PI system make them available to all involved personnel of the refinery, from control room to managers, presenting a unified format with other functionalities available in the refinery information system.

The fact that the on-line optimisation is implemented using a DMC controller, also contributes to the acceptance of the system. Petronor have been using the predictive control technology for many years, and both the technical staff and the control room one are familiar with DMCs.

The DMC controller acts in real-time on the hydrogen network implementing on-line optimisation policies that take also into account the targets provided by the refinery planning and the current state of the proc-ess. The activity of the controller can be monitored and changed on the control room screens by the operators. Priority in the operation is given to maximise the hydrocarbon load, with hydrogen saving coming next. Notice that hydrogen availability can be a limiting factor for hydrocarbon processing or not depending on different factors.

Quantifying the benefits is not easy as it is not possible to measure the use of resources before and after the implementation of a new system in the same conditions. Keeping this in mind, it is possible to perform evaluations of the results and estimate a saving of 2.5% in the hydrogen production, which can represent between 1 and 5 M€/year, while the increment of hydrocarbon loads is more difficult to estimate. In any case, there has been a clear improvement of the operating conditions since the online DMC started functioning, in the sense that hydrogen availability is rarely a bottleneck for production. At the same time, the hydrocarbon production, which is the most valuable target, approached the maximum feasible according to the operating conditions and the targets fixed by the refinery planning, as can be seen in Figure 25. Here optimal total hydrocarbon load and actual one are displayed for a period of nine days of operation, showing that graphs coincide for large intervals as well as the variability of the operation.

Figure 25: Evolution of the optimum total hydrocarbon load (red line), and actual one (green line) for a period of nine days (see document attached)

The DMC optimiser considers in its decisions only the sub-set of plants it operates on, but, in parallel, the RTO implementing data reconciliation and process optimisation takes into account all plants and actuation points. In this way, at a slower pace, it provides a wider view and tools for supervision of the performance of process and the DMC operation, as well as identifying ways in which the operation can be improved.

The overall evaluation of the project results is very positive, as the success criteria initially established had been fully reached:
• A long-term stable solution for resource efficiency monitoring and operator guidance for individual units and overall sites has been implemented, using data reconciliation and REIs.
• A decision support system that uses the computed REIs and generate recommendations to operate at better operating points has been developed and implemented using model based optimisation.
• Roll out across the enterprise has been achieved integrating these functionalities into the refinery information system (PI).
• In addition, the on-line implementation of the optimal operation policies has been made using DMC controllers.

Overall, the system implemented in the refinery is a clear improvement in the efficiency of the use of resources and represents a significant step forward to further integration with other advanced systems in the refinery and enhancements of its functionality.

As an additional benefit, the identification of gaps between the targets given by the planning system of the refinery and what the RTO/DMC compute as feasible targets according to the current condition is a valuable information towards the implementation of feedback to the upper planning layer leading to better tuning and improvements of models in the planning layer.

Most oil refineries have a hydrogen network and plants that consume it for desulphurization of hydrocar-bons or for hydro-treatment, so that, in principle, the results obtained in the Petronor refinery can be ex-tended to other sites following the methodology and using similar tools to the ones utilised in the MORE project. The generalisation will be easy with the Repsol group, but the potential is the same for other com-panies.
Most important obstacles came from:
• The need of a reasonable level of automation and performance in the measurement system.
• The need of a sensible model to support data reconciliation and network optimisation, which re-quires development time and proper software tools.
• The need of personnel with adequate background in modelling and advanced control and optimisation to maintain alive the application over time.


INEOS in Köln operates a typical integrated petrochemical complex, processing mainly naphtha and natural gas as major feedstock to produce a large number of important base chemicals. The site located in Cologne is of high complexity and is characterised by a large number of products and a deep integration of the different plants with respect to products and energy. Large amounts of energy are required for feed conversion and product purification. Only a part of this energy is supplied directly as primary energy (natural gas) or as direct secondary energy (electrical power), a significant portion is generated on site from the by-products of the chemical processes that are mainly used as fuels. The main objectives were to improve the resource management of an industrial site and to achieve significant savings of energy and raw material resources.

INEOS decided to monitor the REIs that are listed in Table 3 out of the REIs that were developed in MORE.

Table 3: REIs selected by INEOS for monitoring purposes (see document attached)

The REIs indicated in table 3 attached were discussed with plant personnel and agreed to be very useful for real-time process operations. Additionally, INEOS has an interest to analyse how much raw material is converted into energy and whether to minimise this stream is economically sensible.

The methods and tools developed in MORE have been implemented on different development levels. The integrated deployment platform was successfully integrated in the INEOS IT infrastructure. The connection to the major information source for process data (PIMS) is implemented and works without restrictions. The visualisation display intranet site is available for all employees with permissions within the office network.

For the site as a whole, a prototype of the site dashboard that is based on the developed aggregation and contribution analysis concepts exists and is frequently used for reporting. The calculation and visualisation is currently run using an engineering tool (MATLAB). The full implementation within the integrated deploy-ment platform is ongoing and will be realised after the end of the project. The site dashboard visualisation is currently only available to few employees. It uses the contribution analysis approach developed within the MORE project to identify the root causes of possible performance deviations. The concept was described as very helpful as it provides additional information to the users which empower them to directly analyse performance deviations.

Full implementations of the REI visualisations for the isoamylene plant and for the AN plant have been fin-ished (s. Figure 26 – the AN dashboard). The data acquisition and visualisation is performed by the inte-grated deployment platform. Different resource efficiency indicators are displayed in parallel which are energy, raw materials or utility related. Energy efficiency indicators are aggregated from the unit-level over the plant level to the plant complex level and provide consistent information on each hierarchical level. The best-achievable practise of the plant was evaluated based on historical data and takes into account the most important performance influencing factor, the plant load. For each production unit, an individual BDP baseline was developed; consequently, the baseline of the higher aggregation level is derived from the performance curves of the lower levels. The performance curve on the highest-level results from four-unit level baselines and provides more reliable performance evaluations compared to a curve that was computed based only upon data from the plant as a whole.

Figure 26: Part of the REI dashboard for the AN plant. Baselines of the plant units are fluctuating with the plant load (see document attached).

The impact on the AN and isoamylene plants is mainly related to a deeper process understanding and a higher awareness of plant personnel regarding resource efficiency. For AN, a potential analysis of the indi-vidual units revealed a saving potential of up to 5% only by operating closer to the best demonstrated prac-tice
The resource efficiency of the operation of the cooling towers is evaluated using a new cooling tower model that considers the ambient conditions. A full implementation of an operator advisory prototype is available for one cooling tower (s. Figure 27). This tool provides information about the operation in the past and the current optimal number of cooling cells to maintain a specified cooling water temperature.

Figure 27: Part of the cooling tower dashboard prototype for O14, running with live data and a weather forecast. (see document attached)

Additionally, the tool uses the latest weather forecast to predict the necessary number of cooling tower cells in the future to maintain the temperature bound and provides information concerning the cooling water temperature in case the number of cells remains constant. The minimum cooling water temperature when running all available cells is also visible to the operators to foresee possible temperature limitations for subsequent processes early. For planning of maintenance activities, the tool can be used to evaluate their influence on the cooling water temperature due to a reduced number of cells. Based on the weather forecast, the optimal point in time for ventilator maintenance can be found.

For the cooling towers, an energy saving of up to 5% is possible by the implementation of the MORE decision support tool. Due to the low temperatures required by the plants that are connected to the prototype cooling tower, the estimated energy saving potential here is 2.5%. However, the evaluation of the impact on cooling water specific energy consumption alone might be misleading as the necessary energy demands strongly depend on ambient conditions, a warm year is more energy intense compared to a cold year.

For a further optimisation of the operation of the crackers, information on the raw material feed (naphtha) is required. Therefore, in the MORE project, spectroscopic analytics for the measurement of the composition of the naphtha stream were developed. Raman spectroscopy based analytics for the cracker were implemented and field tested for eight weeks. The cracker feed was analysed and the results were archived in the INEOS PIMS system.

The impact of the concentrations of the components in the naphtha feed stream on the cracker performance has been evaluated based on the available cracker model. The resulting ranking of the components was used by S-PACT to adapt the analytic prediction model.

The cracker decision support based on these new analytics was tested offline and showed a significant economic improvement potential for its real-time application.

The use of real-time data from plant measurements for REI calculation and data reconciliation has been implemented in the AIMMS cracker model for one of the crackers. Reconciled REIs can be generated based on measured, reconciled or predicted data for the plant and plant sections. The evaluation of historic data for the time period of the analytic field test in figure 28 shows the increase in specific energy consumption after feed quality changes and the slow improvement afterwards as the optimisation has not yet been implemented.

Figure 28: REIORE for cracker complex during feed changes (see document attached)

By applying the advanced analytics together with an optimised operation based on models to the crackers, it has been estimated that an average increase of the marginal gross margin by 1.5% can be achieved. This economic improvement is connected to a decrease of the REIs for raw material use by 0.4% on the average because of the more efficient use of hydrocarbon feedstock while the specific energy consumption does not change significantly. Given the maturity of the process and the successful implementation of debottlenecking projects in the past, this impact is beyond expectations.

INEOS has achieved the following results:
1. Correct evaluation of resource efficiency in chemical plants with heat generating or consuming reactions in one integrated REI showing energy and material resource efficiency
2. Site wide structuring of the plant hierarchy and site wide bottom-up aggregation of generic REIs
3. Site wide roll-out of MORE dashboards has started.
4. Methods for the computation of the best operating points based on historical data as baselines for the REIs.
5. Implementation of the integrated deployment platform
6. Implementation of real-time analytics for the cracker feed.

The success criteria for INEOS were fulfilled as appropriate for an R&D project, the implementation on the site is ongoing.

In 2017, the goal is to integrate each plant into the site dashboard with the corresponding BDP curves and to generate a comprehensive real-time dashboard which visualises the current performance against the best possible performance and indicates the performance gaps due to sub-optimal operation or other relevant influencing factors.

An investment decision for a permanent installation of the analytics will be made in 2017.

In 2018, the plants which were not used for the development of MORE prototypes will be analysed in detail and decomposed into sub-plants to achieve more meaningful performance indicators and BDP models.

Regarding the cracker model, the business case for the online analytics will be evaluated and a permanent online measurement will probably be realised in late 2017/2018.

INEOS will pursue the objectives with follow-up internal projects to fully achieve all criteria by the end of 2018.


The BASF Personal Care and Nutrition GmbH production site in Düsseldorf, Germany, comprises a continu-ously operating upstream section and a number of batch processing plants that produce highly specific products. The value chain is based on natural and renewable raw materials.
The case study for MORE was selected according to the following criteria:
An integrated batch and continuous production of reasonable complexity with online production data con-tinuously recorded, clearly defined system boundaries, external influences (e.g. raw material variations, partial load).
The selected case study (Figure29) fulfils these requirements.
The case study shows specific challenges: coupled batch and continuous operation, significant number of operator decisions that influence the efficiency of the operation, changes of feedstock because of the use of plant oils of different origin.

Figure 29: Process scheme of the BASF PCN case study (see document attached)

The chosen case study is interconnected by streams of intermediates, products, water of different purity levels, carriers of energy and waste streams. Its current mode of operation aims at reaching the target specification at all conditions of fluctuating raw material qualities, the so-called “one-size-fits-all” approach. Selected resource efficiency indicators are monitored online and sustainability indicators are evaluated offline. They are used for process optimisation projects, but they are not used for the real-time steering of the process.

The indicators that have been developed during the MORE project are oriented towards indicating material efficiency, energy efficiency, environmental load, and product quality. In detail, there are indicators for the consumption of raw materials, energy consumption, water discharge, emission and solid waste. Pursuing the aim of collecting enough online data to monitor these REIs, different online analytical approaches were tested during the project. The computation and visualisation of the REIs was focused on the most promising steps of the BASF case, the purification and post-processing operations (Figure 30). For these processing steps, REIs for material- and energy consumption were monitored online.

As a first step, the dashboard was provided to the process engineers. The objective was the verification of the data and feedback to the first version of the dashboard. During the implementation and validation of the dashboard, a rented UV sensor was used to measure a crucial quality parameter of the final product. This additional information enables the operators to identify and to respond to changes of the process without delays. The UV sensor was tested and validated during two production campaigns.

Figure 30: REI-Dashboard of the case study of BASF (see document attached)

The dashboard includes modern HMI elements, e.g. bullet charts. A bullet chart shows in one view and in an intuitively comprehensible manner:
1. The actual REI values
2. The variance of the REI values in the last time period
3. The current tendency of the REI values
4. The alarm limit linked to the REI.

With the goal to provide information that can be used to improve the resource efficiency of the operation of the plant, several PAT approaches were identified and tested.

The aim of all approaches was to get in-depth insights into the process and thus a possibility to monitor REIs at the reaction, purification and post processing step. The most promising leverage was predicted for the post processing part using UV-Vis spectroscopy. This new analytical online method is a promising alternative to the currently used offline analysis. The feasibility of the online analytics was proven using a rented instrument. Potential savings could be evaluated during two campaigns under real processing conditions. The online signal was directly visualised on the dashboard beside other REIs for the post processing step. The calibration of the processing of the online signal was conducted during a first campaign with offline measurements and the first model was improved. The second campaign was used for the evaluation of possible savings at plant level. The result of the two campaigns was that the online signal permits a faster response to process changes than offline lab measurements. Especially the start-up procedure can be optimised with better insights into the process. A transfer of the knowledge gained to other products is promising, due to the universal analysis method of the UV-Vis measurement. Results based on the two campaigns with the rental equipment indicate that the use of bleaching agent can be reduced by around 2 %, if the new online equipment is implemented. A possible investment is promising due to the reduced consumption of bleaching agent and reduced labour hours. The currently used offline analytics are time consuming and support the operators only at a low frequency.


The case studies that were considered in the MORE project cover a significant part of the value chain of the chemical industry. Raw materials of the companies providing the case studies are both fossil and renewable feedstock. The processes are equipped with standard measurements that can be used for the calculation of REIs and show potential for the use of advanced methods and tools such as model-based data reconciliation and implementation of new PAT sensors. In two of the case studies the flows of interest contain many chemical species so that it is justified to use advanced online analytics to calculate the REIs and to optimise the processes based on these additional measurements. All cases show a large potential for the use and the exploitation of REIs and real-time optimisation, and – most importantly – in all cases the support of the plant personnel support was gained.

Overall, all technologies that were developed in MORE have been successfully applied in at least one of the industrial case studies and the implementation of the tools and methods is still ongoing with high expecta-tions and a strong commitment of the industrial partners. Two partners, Petronor and Lenzing implemented control solutions that are in daily use and have a direct impact on the performance of the plant. Further, model-based decision support tools have been developed for these two cases and are in use (Petronor) or under field test (Lenzing). The technologies can be applied to other companies that operate the same type of plants. The two other partners, INEOS and BASF, installed prototypes of the MORE technology on a scale that matches the size of the case study. INEOS is committed to follow through with the full installation of the MORE technology for REI visualisation, real-time reporting and decision support.

The SME partner LeiKon has productised the MORE implementation platform and is involved in further projects that are part of the implementation and roll-out of the MORE technologies at INEOS. As far as possible for confidentiality reasons, the results of MORE have been published in scientific papers and are a core element in the upcoming book on “Resource Efficiency of Processing Plants – Monitoring and Improvement” which is edited by Stefan Krämer and Sebastian Engell of the MORE consortium and which will be published by Wiley-VCH in 2017.

Potential Impact:
1. Impact

The impact of applying the MORE approach in the four industrial companies (the plants of the “case stud-ies”) involved in the project was assessed and the potential impact of generalising this approach to other plants or companies on a European level.

The purpose of the impact assessment was to evaluate the economic, environmental and managerial impact of using REIs in the chemical and process industries for four case studies, a refinery (Petronor, Spain), a petrochemical site (INEOS in Köln, Germany), a global supplier of specialty chemical products and nutritional ingredients based on renewable raw materials (BASF PCN, Germany) and a plant producing viscose fibers (Lenzing AG, Austria) (see fig. 31). The generalisation analysis estimated the potential impact of the broad application of the MORE approach in Europe.

Figure 31: Objectives of the impact assessment (see document attached)

The assessment was done in four complementary dimensions:
- Analysis of the environmental impacts in the four industrial cases that implemented MORE ap-proaches during the project. The assessment used the life cycle analysis methodology and compared the environmental performance with respect to relevant factors (reduction of CO2, reduction of energy used, etc.) before and after the implementation of MORE;
- Analysis of the economic impacts in the four cases as observed during the course of the project or calculated from the reduction potential of production costs.;
- Analysis of changes of human decision making induced by the implementation and the use of REIs in the companies at the various levels of decisions (plant operation, plant management, strategic management);
- Generalisation of the MORE results to estimate to which extent significant impacts can be reached by implementing the MORE approach on a wider range in the European chemical industry or other sectors of the process industry.

The analysis of the impact of the four MORE case studies and of the generalisation potential was placed in the context of the overall European process industries and more specifically the chemical industry.

The four cases represent a good coverage of the value chain of this very diverse industry sector. However, even though the MORE cases can be considered as “standard” examples of companies of the chemical industry, it should be noted that the application of resource efficiency measures in industrial processes can take place in quite different situations: chemical companies in Europe operate on different levels with regard to instrumentation, automation, operator support, management support, and real-time optimisation., Some companies and plants have already applied optimisation and model-based control of plant operations whereas others are only using conventional control and there are also plants with a low level of instrumentations and a large amount of direct human intervention in the production. This means that the generalisation potential needs to be taken with care: it is assumed in our analysis that 50% of the chemical companies in Europe can take up the resource efficiency monitoring and improvement methods from MORE directly, while in the remaining cases either much of the potential has already been realised or the state of plant instrumentation and automation is insufficient for their application in the short term and first some modernisation of the plants is needed.

Impact assessment:

The impact assessment methodology combined several tools to measure the environmental impact assessment and the economic impact assessment: The environmental impact assessment was conducted based on the Life Cycle Assessment (LCA) methodology, while the economic impact assessment combined quantitative and qualitative approaches to measure the gains of optimisation and the changes resulting from the project. In terms of the approach to data collection, environmental and economic impacts were assessed by comparing the situation “before” and “after” the REIs had been defined and displayed to operators or used in model-based optimisation and control schemes.

Main findings of the impact assessment:
The particularity of the MORE project is that it not only focused on developing new methods, techniques and tools but was really connected to the daily work in the chemical and process industries. The REIs developed in the project concern the different hierarchical levels of the industrial companies from operating levels to business and strategic levels. These levels have been considered in the analysis of the impact as the expectations (and thus the expected impacts) have various dimensions.

We identified five levels of impacts. Three levels are inside the companies, where we distinguish the operating level, the managing level and the strategic level. In addition, there also is an interest to understand to what extent the REIs can be spread at European level to other comparable companies, and how the analysis can be extended to the entire value chain, including upstream and downstream processes (e.g. raw material acquisition, energy production and use phase).

The diagram (Figure 32 in the document attached) represents the logical framework of the project, linking the “activities”, performed in the project (the definition, computation and utilization of the REIs) and the expected impacts as they have been identified by the industrials partners.

Figure 32: Logical framework of the use of REIs (see document attached)

The diagram emphasises the different levels of impacts: Plant level (operating and managing level), company level (strategic level), and finally sector level. The analysis focused on plant and company levels, starting from the use cases, and was then extrapolated to the sector level.

The assessment results for each of the companies in the consortium are summarised below:
1. For Petronor (Muskiz, Spain, Repsol group), the second biggest oil refinery in Spain, the economic gain from the implementation of MORE is estimated to be between 1.000.000 to 5.000.000 €/y which corre-sponds to the target of 3-5% of savings of the cost of hydrogen production. It also had an impact on re-ducing greenhouse gas emissions from the hydrogen production by about 3.5%, as for the same amount of hydrotreated diesel less hydrogen is needed. Regarding the impacts on operational and management decisions, the implementation of the REIs impacts the company mainly at the operational level by taking profit of the new tools developed in the MORE project. Opportunities to spread the approach in other plants of the company have been identified and can be achieved within a 2-year horizon.
2. The integrated petrochemical site of INEOS in Cologne, Germany is a major producer of base chemicals. In this case, the implementation of the MORE approach is still under way and will continue after the end of the project, especially regarding the implementation of new dashboards for the visualisation of REIs in different plants. Only an estimation of impacts can therefore be given at this stage. The example shows a considerable potential for economic and environmental improvements. Regarding the environmental impact, the reduction in indirect greenhouse gas emissions (through electricity savings) could reach 16,400-32,900 t CO2 eq less emitted per year. Beyond the MORE project, INEOS has a strong interest in investing further in energy efficiency improvement measures and in spreading the results of the MORE project more widely in the company.
3. BASF PCN is a global supplier of specialty chemical products and nutritional ingredients based on renewable raw materials. The definition and implementation of REIs did not yet lead to the expected results, due to technical difficulties encountered. The implementation of REIs can produce potential economic impacts, but further investigations and investments are required. Regarding the environmental footprint, the impact so far is small as the MORE technology could be applied only to a small part of the process under consideration. However, the gain in knowledge provides a basis to continue the investigation in other plants.
4. The Lenzing site is a reference factory around the world for producing man-made cellulose fibres. Overall, the economic impact of all optimisation measures using MORE approaches in savings of steam consumption could lead to economic benefits of 575,000 to 825,000 € per year. There is coupled to a significant effect on climate change, by reducing direct CO2 emissions from the site by about 0.3%. Further investments will be made beyond the project in energy efficiency and it is planned to spread the MORE approach in other plants of the company (in China and Indonesia) and in others application areas, increasing the savings.


The objective of the MORE generalisation analysis was to find out to which extent significant impacts can be reached by implementing the MORE approach on a wider level in the European chemical industry and in other sectors of the process industry. Regarding the generalisation methodology, the generalisation from the data gathered during the Impact Assessment was done along three axes: generalisation to comparable processes or products, generalisation with respect to comparable equipment or technologies, and generalisation of the general approach of MORE to the process industry.

The methodology was based on a five step approach: desk research on relevant background data on chemical processes and the process industry in Europe (e.g. production capacities, etc.), set up of hypotheses on how the MORE impact assessment of the case studies can be cast into general terms, so that it can be generalised to a broader range of cases, collection of real impact data from the “after the project” analysis, and application of the generalisation to other plants or other sites within the same company, or to other companies on European level.

Main findings of the generalisation:

For PETRONOR, on the basis of the available impact assessment data, two generalisation hypotheses could be applied on the internal level (the Repsol Group) and on European level, one with regard to energy re-sources and environmental impact, one with regard to economic benefits:
1. Increased diesel production with the same amount of hydrogen: if this was applied to all EU refineries the potential production increase was 2.8-5.6M tons of Diesel fuel per year in the EU.
2. Cost saving: the generalisation potential in economic terms is at least 25-125 M€/y and at best 50-250M€/y in the EU.

For INEOS: Within the MORE project several plants have been in the focus for which different technologies of MORE have been applied: an AN (Acrylonitrile) plant, a cracker and cooling towers. INEOS in Köln being a highly integrated site, the impact assessment data could only partly be translated into generalisation figures. The generalisation analysis has shown that at least 25% of the large scale continuous base chemical plants in Europe could reach an overall energy efficiency improvement of 2% by fully implementing the MORE methods as operator advisory systems. INEOS estimates an improvement potential on site of 1.5% variable gross margin increase in their crackers, a potential 2.5% energy savings in their cooling towers and a 1.25-2.5% energy savings in their AN plants.

For BASF: Specialty chemicals are produced usually in small volumes but represent 28% of the European chemical sales. Quantitative impacts on European level are difficult to estimate and the uncertainty is too high for giving recommendations and for estimates on possible savings and efficiency improvements.

For LENZING: The generalisation analysis has shown that there is potential for impact beyond the individual plants addressed in the MORE project on two levels:
1. Energy and steam saving: Due to an optimised process, less energy (fossil fuels) is needed for the same amount of product. This results in a saving of steam in the evaporation process, thus leading to natural gas savings and to cost and CO2 emission savings. The optimisation potential amounts to about 36-72kt/year and potential energy savings of 2.5-5 MJ/year, as well as 0.35-0.7Million Nm3 natural gas sav-ing per year in Europe.
2. On the basis of the reductions in energy and steam consumption displayed above, the savings potential amounts to 23.5K-47K€/year for smaller structures in Europe.

As a conclusion, there is a significant potential for improving resource efficiency in terms of materials and energy consumption and financial costs, both within the companies involved in the MORE project and as-sessed in the case studies, and in similar companies within the same sectors across Europe.

Several recommendations are provided to policymakers in order to support the process of wider implementation of resource efficiency solutions in the European process industry:
• The human operators have and will have for a long period of time a significant influence on the op-eration of plants in the process industries. Therefore, the support and the acceptance of the intro-duction of new tools and technologies by the operators are key to success. Collaboration between human operators and computer-based tools should be addressed prominently in further research projects.
• While the technology level in large chemical plants generally is high and many opportunities for re-source efficiency improvement have already been seized, many smaller plants exist which have not reached this level. As the engineering workforce in such plants usually is small, external support is needed and measures should first address low hanging fruit.
• Digitalisation of industry is a prominent topic in research and innovation at the moment. In MORE, we have encountered that the lack of reliable and precise measurements of flows of materials and of energy in the process industries which is a basic ingredient for digitalisation and is a major obstacle to the use of the available data.
• It was noticed that data about production data of the chemical industry is dispersed over many publications and difficult to access, so the development of reference publications is encouraged.
• Further progress in energy and resource efficiency requires significant investments. Such invest-ments will only be made if there is a stable regulatory framework. Conditions are needed so that the risk that is incurred for longer payback periods is reduced. Otherwise, the potential of the available technologies cannot be realised.

2 Dissemination and Exploitation

Industrial Stakeholder Panel

The MORE consortium has set up an Industrial Stakeholder Panel (ISP). Members of this panel were nine external experts from various sectors of the European process industries such as pharmaceuticals, chemicals, refining, pulp and paper, steel and consumer products and representatives of the four MORE industrial partners.

In order to extend the perimeter of the outreach activities and to get a multiplying effect, the stakeholder panel consisted mostly of responsible engineers and managers of producing companies who could provide practical assessments and ideas.

The panel members advised the MORE consortium with respect to promising research directions and possible dissemination activities from both, the industrial and the technological point of view, provide comments and suggestions on possible impact and exploitation.

The external members of the ISP were involved as active observers in the project activities. They provided input on the definition and application of the novel real-time resource efficiency indicators in different in-dustrial sectors and actively promoted the future take-up of the results in the European process industry. The ISP met three times during the life-time of the project; the last meeting was integrated into the Final Public Workshop on the results of the project.

The members of the panel were informed about the progress of MORE and were given access to the infor-mation package of MORE. The outreach to these stakeholders and decision makers took place in parallel to the project development, right from its beginning, as these players were vital to any further and lasting use of the technologies, whether in science or in industry.

The identified REIs, the first developments in real-time monitoring and optimisation of resource efficiency in integrated processing plants and their transferability to other sectors were discussed, aiming at making the project results suitable for implementation in a broad range of sectors of the process industries. The ISP members provided inputs on the vision and needs of their companies and sectors.

At the MORE final event in Frankfurt on February 15-16, 2017, a panel with experts from the ISP and other stakeholders discussed and confirmed the applicability of the MORE approach in other industries.

Figure 33: Photo of the Panel with ISP members at the MORE final event (see document attached)

Guidebook to the definition and implementation of resource efficiency indicators

A guidebook has been created to support the European process industry in the identification and selection of real-time REIs that are relevant from the operational point of view. This step-by-step guidance was prepared in close co-operation of the project partners VTT, INEOS, TUDO, LeiKon and UVA utilising the experience gained during the implementation of MORE indicators and decision support in the industrial case studies as well as in pilot studies for two other industries, pulp and paper and sugar industry. The methodology consists of twelve steps for selecting and defining the process units for consideration, identifying and selecting REIs, and implementing and evaluating them. The guidebook provides a technical introduction to the methods that should be used and a detailed step-by-step procedure that can be followed to successfully implement the core ideas of real-time resource efficiency monitoring and improvement. For reference, additional technical details are provided in the appendices.

The guidebook is available via the MORE website in several versions:
• Deliverable D6.6. Step-by-step guidebook on “How to succeed in the identification and implementation of real-time or near real-time Resource Efficiency Indicators for your plant or industrial sector” (Deliverables Section)
• Condensed, nicely formatted version (brochure, 48 pages, published by INEOS, available here) and
• Full version (scientific publication, 78 pages) that appeared in the VTT publications (available here; both also in the Publication Repository Section).

The systematic concept for a successful selection and implementation process of REIs, as described in the Guidebook was presented at the MORE Final Event at DECHEMA on February 15th-16th, 2017 by providing the 48-page brochure in print to the attendees, by the presentation “Improving resource efficiency – a step by step guide” by Tiina Pajula (VTT) and by a poster in the poster session presented by Juha Hakala (Senior Scientist, VTT).

SPIRE workshop “Resource Efficiency Monitoring, Assessment and Optimisation”, Brussels, January 2016

The MORE consortium initiated the SPIRE workshop on Resource Efficiency Monitoring, Assessment and Optimisation”, which was organised by MORE with the support of A.SPIRE in cooperation with the three ongoing EU-projects MORE, TOP-REF and REFFIBRE, that were funded under the same call. The workshop took place on 27 January 2016. The workshop was designed for representatives from the ongoing SPIRE and FP7 projects, as well as for possible future project proposers, aiming at sharing in-depth information about the results that had been obtained in the on-going projects, hence avoiding duplication of efforts and stimulating new research and innovation.

The program was organised in three thematic sessions with technical presentations, and included an interactive round table discussion. Participants learnt about the approach and innovations developed and implemented in the three ongoing EU-projects MORE, TOP-REF and REFFIBRE that are addressing the topic of Monitoring, Assessing and Improving Resource-efficiency in the Value Chain of the Process Industries, and exchanged on technical issues and discussed new ideas that are relevant to resource efficiency assessment and improvement in the process industries in Europe.

MORE final event « Real-time Monitoring and Optimisation of Resource Efficiency – From Measure-ments to Optimal Operation”, Frankfurt, February 15-16, 2017
After more than three years of successful implementation of the MORE project, a Final Project Workshop on “Real-time Monitoring and Optimisation of Resource Efficiency – From Measurements to Optimal Operation” was held on 15-16 February 2017 at the premises of DECHEMA in Frankfurt, Germany.

The aim of the workshop was to share the results and experiences and to discuss them with a broad audi-ence of experts from different sectors of the process industries.
During the workshop, more than 60 participants attentively listened to presentations about the outcomes of the project and discussed the developments in monitoring and optimisation of resource-efficiency in the value chain of the process industries.

At the welcome session of the workshop, Ivan Scannapiecoro, the European Commission felicitated the project for its success and impact and stressed the importance of the results for other process industries, and Àngels Orduña, the Executive Director of A. SPIRE, outlined the importance of resource efficiency for the process industries in Europe and the contribution of the SPIRE program within Horizon 2020 and the role of the SPIRE Association.

The scientific programme included, among others, a presentation of the MORE Resource Efficiency Indica-tors (REIs), and presentations of different results of the case studies of the MORE project.

A panel with well-known experts from the European process industries moderated by Prof. Engell (TU Dortmund, Scientific Leader of MORE) discussed the applicability of the MORE approach in other companies and industries, the lessons learnt, challenges, and steps forward towards a more sustainable production in Europe. Panelists were Dr. Kai Dadhe (Evonik), Dr. Alex van Delft (DSM), Dr. Martin Jenke (CEMEX), Dr. Stefan Krämer (INEOS in Köln), Dr. Günther Windecker (BASF) and Dr. Martin Winter (CEFIC and A.SPIRE).

The experts agreed that the developed resource efficiency indicators and methods for optimisation and decision support were indeed “relevant for the diversity of processes we have in the company”, as stated by Alex van Delft, DSM. He added that he was “very pleased to see also that the “people” aspect has rightly been taken into account, as engaging staff (operators) is very important.“ Kai Dadhe, Evonik, added, with the agreement of the whole panel that “the visualisation tools are specifically convincing – and you need these when you want to show to the operators where potential improvements can be made. (...) At Evonik we are also interested in evaluating the feasibility of applying the technology developed by the MORE project partner LeiKon in our system.”

Further topics of the discussion concerned aspects of IT integration, long-term maintenance of solutions, and pressing topics for future research and innovation projects. Martin Winter announced that the aspect of digitalisation will play an important role in future SPIRE calls as significant improvements can be achieved without large investments into new plants.

The participants continued the discussion in small groups after the end of the plenary discussion, at a poster session with refreshments. The project findings were comprehensively summarised in 11 posters and in combination with a live presentation of the deployment platform “Intexc Suite”, that was developed by LeiKon during the MORE project, as well as the work of S-PACT on analytics, the developed methodologies and application cases were presented to an interested audience.

The final workshop confirmed that the MORE project followed the right approach and will have a significant impact, both in the companies involved and on a broader scale.

Other dissemination activities

The dissemination activities in MORE were based on the Communication and Dissemination Strategy elabo-rated and issued at the early project stage. The following main actions can be enumerated:
• The project branding (project logo, visual identity, templates) has been defined during the first 3 months of the project.
• The elaboration of further dissemination material (flyer, poster) was made within the first year of the project (2014), the target groups were identified, and the communication activities organised.
• The project website was designed, set up and put online in January 2014 and has been maintained and continuously updated; related events, activities and news were announced on the web portal and reports published. There were 7.700+ visits of the MORE website since its creation and close to 20.000 page views. The project website will be maintained during at least 2 years after the project end.
• Publication of 21 articles in peer-reviewed publications (journals and conference proceedings) by the MORE team. The book “Improving Resource Efficiency in Processing Plants”, edited by Stefan Krämer, INEOS and Sebastian Engell, TUDO, the scientific and industrial leaders of MORE which will be released by the end of 2017 by Wiley-VCH contains, among other contributions, the main results of MORE as a core element.
• Additionally, the project was invited for interviews that were – among others - published in a num-ber of journals, e.g. the Pan European Network journal, the Process press or within the European Commission website in several publications (success story June 2016, January 2017.
• All in all, MORE has been presented in over 40 scientific events, where publicity material and presentations have also been distributed. Besides these, MORE team members presented the project in a number of other events, for example, in March 2016, Prof. Sebastian Engell (TU Dortmund), Scientific Leader of the MORE Project and Consortium Member Benedikt Beisheim (INEOS in Köln) participated in the ener.CON Europe conference in Berlin where they moderated a roundtable with mostly industry participants. A “Workshop on Resource Efficiency Monitoring, Assessment and Optimisation”, organised by SPIRE in cooperation with MORE, with presentations from its “sister projects” TOP-REF and REFFIBRE, took place on 27 January 2016 with several presentations from MORE. Stefan Krämer and Sebastian Engell gave an invited one hour keynote speech at an industrial workshop of the German engineering company plantIng in Cologne on September 8, 2016 that was followed by long formal and informal discussions. Sebastian Engell gave an invited plenary lecture at the 1st IFAC Symposium on Cyber-physical and Human Systems in Florianopolis, Brazil, on Dec. 8, 2016 on “Operator Support for Improving Resource Efficiency in Chemical Plants that presented the achievements of MORE. Further exchange has taken place specifically on standardisation topics, e.g. with at the NAMUR Annual General Meetings (see further information in the chapter on Standardisation activities below).


The results of the MORE project are exploitable on different levels: scientifically, commercially and also pre-commercially. An exploitation plan was set-up and continuously updated which includes strategies to exploit the project results of MORE on the international market. The exploitation plan was specified for the consortium as a whole and for each partner individually. The exploitation activities were closely coupled with the MORE communication strategy and communication tools as well with the standardisation activities, Industrial Stakeholder Panel activities and the impact assessment. The objectives of the exploitation efforts are as shown in Figure 34.

Figure 34: Exploitation objectives (see document attached)

In MORE three different aspects of exploitation goals were pursued (s. Figure 35). Scientific and partly also pre-commercial exploitation results were mainly targeted by the partners TUDO, UVA, inno and VTT. Com-mercial exploitation results were pursued by the service and tool providers S-PACT and LeiKon as well as by the end users INEOS, BASF, Lenzing and Petronor. Pre-commercial activities were done by all partners, standardisation mainly by the partners VTT and LeiKon.

Figure 35: Classification of exploitation goals (see document attached)

Publications are the most common way to show and to share scientific results. A publication repository was maintained. The list of scientific papers published is available. One main exploitation result of the consor-tium as a whole is the Step-by-Step Guidebook published as publicly available booklet (Condensed, nicely layouted version (available here) and full version (scientific publication, 78 pages) that appeared in the VTT publications (available here; both also in the Publication Repository Section). Besides publications in scien-tific journals and conference volumes, the methods developed in MORE will also be published in an edited book “Improving Resource Efficiency in Processing Plants”. The editors are Stefan Krämer, INEOS and Sebastian Engell, TUDO, the scientific and industrial leaders of MORE. The book will be released by the end of 2017 by Wiley-VCH. It contains, among other contributions, the main results of MORE as a core element.

A large part of the MORE project work was oriented towards industrial application. Consequently, industrial exploitation of the MORE results assumed a major role in the exploitation strategy. The end users exploit the MORE results directly to improve their operations, from REI visualisation to improved control, decision support and online optimisation, as well as for awareness and extension of ISO 50001:2011, ISO 50006:2016 and ISO 14001:2015 activities. All end users plan to further use the tools that were developed in MORE in order to run resource efficient and environmentally friendly production plants. Project results will be further developed and rolled out to other technologies and production sites. The tool providers LeiKon and S-PACT exploit the MORE results for new business strategies based on licence fees as well as on consulting and engineering services.

Beside planning and consulting activities, LeiKon launched a new successful business segment “LeiKon solutions and services” in 2010. LeiKon integrated the new innovative deployment platform in the business strategy of this growing segment. In 2014 LeiKon started the branding of the new Deployment Platform. The brand name “IntexcSuite” (Intexc =Integrational Excellence) was established and a graphic designer designed a logo for the tool. Based on these first marketing activities further activities to increase a corporate identity were initiated. Beyond the focus of REI applications, IntexcSuite will be placed as a developing framework to implement production assistant systems. Production assistant systems shall help operators and plant engineers to control and monitor production processes in an advanced manner (REIs, prediction, decision support, etc.).
S-PACT exploitation was focusing on three main project outcomes:
1. Concepts and tools supporting the calibration workflow for spectroscopy
2. Business cooperation with Process Industries and equipment manufacturers
3. Application Cases in hydrocarbon and renewables processing.

The method for selecting the most relevant features for the measurement task supports the process of sample selection from historical material data and allows to streamline the “retro-DoE” approach (DoE: Design of Experiments). The “retro-DoE” sample selection method has been implemented prototypically and is continuously being validated in customer projects outside MORE. It is expected to support future S-PACT calibration development by more efficient workflows. The development of the online cracker feed analyser was performed on a Raman spectrometer system provided by Kaiser Optical Systems (an Endress+Hauser company). The success of the approach provided by S-PACT has raised very strong interest at Kaiser and E+H to intensify collaboration with S-PACT in the field of application development through chemometric modelling, not only for the application field in MORE. Since E+H is heavily strengthening the “Advanced Analytics” sector (i.e. beyond physical sensors like pressure, temperature, pH, mass flow, level control) through integration of companies like Kaiser, future collaboration between S-PACT and E+H is expected to grow S-PACT business also.


With respect to standardisation, the following activities were pursued:
• Screening existing standards and on-going standardisation activities to identify actions that are re-lated to energy and resource efficiency of process industries and are relevant to the MORE project
• Attracting attention from national and international professional societies and standardisation bod-ies to the proposed real-time resource efficiency indicators
• Communicating the results in an early phase of the project to national or international professional organisations and standardisation committees
• Approach national or international standardisation committees to integrate real-time resource effi-ciency indicators into o integrate REI definitions into recommendations or maintenance requests of normative standards
• Ensuring the compatibility of MORE REIs and REI methodologies with relevant existing standards.
• This general approach was mapped into a specific strategy which is represented graphically in figure 36.

Figure 36: Standardization Roadmap realized in MORE (see document attached)

Based on the screening of existing standards new contributions of MORE which are suitable as inputs to standards were identified. The result was that the definitions of REIs as well as the methodology to identify and to realise REI applications were identified as main issues worth to standardise. In order to attract attention from national and international professional societies and standardisation bodies to the results of MORE a number of standardisation meetings have been attended and contacts established during the project. As a result, three standardisation bodies were motivated to initiate new standards or to integrate the MORE results into ongoing standardisation activities: NAMUR and IEC supported by DIN DKE (D6.9). It was agreed with the NAMUR managing board to establish a new NAMUR Ad-hoc Standardisation Group “Resource Efficiency Indicators”. A NAMUR recommendation describing the state-of-the-art and best demonstrated available technologies is planned for 2017. This will serve as an input for further national and international standardisation activities. Additionally, the German DIN DKE K931 agreed to support a new work item proposal for an international standard in terms of an IEC Technical Report (IEC TR) based on the NAMUR recommendation.
List of Websites:
Project public website and relevant contact details

The pre-final version of the MORE website ( was available in January 2014, finalised by end of May 2014 and has been maintained since. It contains a home page with a slider presenting the main goal of the MORE project, the published news on major activities and results and additional pages presenting the project activity in more details, as well as all partners. A public repository contains all public deliverables as well as additional publications.

The MORE website represented the first vehicle in raising awareness of the project and contains a general presentation of the project activities, the consortium, case studies etc. It follows the MORE branding and was updated all along the project lifetime with latest results and findings.

Project activities, news and related information were announced on the web portal, and public re-ports/presentations were published. inno TSD was the overall responsible for the website’s administration and function. The project partners contributed to the content and provided materials to inno TSD. The web site also provides access to public project deliverables and project outcomes.

The website will be maintained during at least 2 years after the project end.

Figure 37: The MORE Project website (see document attached)

The website was designed using responsive web technologies to enable optimum visualisation independ-ently of the targeted device (PC, tablet, and mobile).
There were 7.800+ visits of the MORE website since its creation, and more than 20.100 page views. Some statistical information for the website (covering the entire project period) is shown in the graphs in the figure 38 attached.

Figure 38: MORE project website statistics (see document attached)

Project website:

Project Coordinator:
Svetlana Klessova, inno TSD, France (

Scientific Leader:
Prof. Sebastian Engell, Technische Universität Dortmund, Germany

Industrial Application Coordinator:
Dr.-Ing. Stefan Krämer, INEOS in Köln, Germany