ABSTRACT. In This talk, I will discuss the role of M&S in the new ecosystem of techniques for data analysis; the audience will get some light on how I consider the different techniques should be coupled to solve current real problems in different transport modalities ranging from Road to Aviation. This talk will give also direction to young professionals on what are the key areas to put focus on if they want to pursue a successful career in the future using current techniques like AI, Big Data, Process and Data Mining, optimization, Statistical analysis and of course simulation.

ABSTRACT. Accurate simulation of the numerical optimal control in software environments where call to simulation routines is explicit, for instance Matlab and SciPy. A discussion on the simulation aspects of numerical optimal control, how it may fail, and how such erroneous results can be detected using accurate simulation. The key contribution is how to accurately include a piecewise constant control input in the simulations, which is discussed in detail, including code examples. The technique is demonstrated on an example problem which show how simulation can be used to analyze optimal control problems with uncertainty, but also demonstrates how erroneous simulation may lead to erroneous conclusions.

ABSTRACT. Future climate is expected to be warmer, more humid, and cloudier with more frequent extreme weather conditions. Current building design should consider these changes as they can significantly influence the function of buildings in the future. Here, we study common building envelope assembly subjected to different climatic scenarios. An experiment was set up to validate a numerical model, which is further applied to assess hygrothermal performance (heat and moisture transfer) of the building envelope subjected to different boundary conditions. The assessment is provided via Finnish mould growth model that identifies risk of biological growth through dynamic hygrothermal conditions. Finnish meteorological institute provides data that predicts the climate in 2030, 2050 and 2100. The humidity inside the building envelope is assumed to increase slightly in time, however, increased temperature in the future may cause more favorable conditions for mould growth, especially, if mould sensitive building materials are used. The hygrothermal assessment of building structures with consideration of climate change in structural design is a key factor to provide sustainable building designs. Numerical model was successfully validated with experiments providing data within tolerances of measurement equipment.

ABSTRACT. The operational conditions at the Dalsfoss power station are complicated due to many requirements such as environmental regulations and safety constraints. Model predictive control (MPC) has been in use at this power station to control the floodgates at the Dalsfoss dam. However, the current formulation of MPC at the power plant does not have routines to explicitly handle output constraints. In this paper, a new improved optimal control problem (OCP) is formulated for the operation of the flood gates at the Dalsfoss power station. This new OCP formulation is thought to be relatively easier for the operators to understand and it is more flexible to the violation of constraints. The aim of this paper is to extend the current MPC used at the power plant so that the output constraints are systematically included in the new improved MPC formulation. Two alternatives are presented and their robustness to an uncertain disturbance is analyzed through robustness analysis.

ABSTRACT. 1 Background Bergen Engine AS is a developer and producer of gas and diesel engines for the marine and land based power market. The latest LNG fuel engine developed is the B36:45 engine family and has been in operation since late 2018. It is a medium speed lean-burn single fuel spark ignited internal combustion engine. It is turbocharged and has a 2-stage water cooled charge air cooler. It runs with a high air to fuel ratio compared to the required air for a stoichiometric combustion. This is done to lower the combustion temperature and hence to reduce NOx emissions. This lean mixture of air and fuel is difficult to ignite. A pre-combustion chamber mounted in the cylinder head is used to ignite a rich mixture, and the resulting flames will propagate out and into the the main chamber where it will ignite the lean mixture. The air pressure control is the most influential of all the control loops currently present at the B36:45 engine. It dictates most of the engine behavior as it directly controls the air/fuel ratio under all operational conditions. The set point to the air pressure controller is based on a map with engine power output and engine speed as inputs. The set point map is derived based on numerous test runs at the test facility by skilled engineers. Since this is a static map, the air pressure set points can be biased to a certain degree based on operational conditions. The ignition time is the time in crank angle (CA in degrees) at which the cylinder individual spark plug is ignited in the pre combustion chamber. Each individual cylinder can adjust its ignition timing by ± 3 degCA from its base value in order to adjust the power output and peak pressure for each cylinder. Obtaining and adjusting the base value for the ignition timing, and control of timing location is complex and difficult to maintain. 2 Aims LNG fuel internal combustion engines are very complex systems. In general, it is difficult to use the first principles methods (conservation of mass, energy, momentum etc.) to model such a system, and they usually yield very complex dynamic models. In the latest B36:45 engines, the instrumentation and diagnostic capabilities has increased significantly due to the possibility to perform in-cylinder pressure monitoring for each cylinder on cycle-to-cycle basis. This gives a large amount of valuable measured data which can be utilized further. One of the aim of the paper is to use these measured/logged data in order to develop a data driven model capable of showing some of the important dynamics of this engine. The developed model is further used to make an advanced model based controller for increased efficiency of the B36:45 LNG engine. One of the goals is to generate a dynamic setpoints (instead of using static maps) for charged air pressure which are optimal for maximizing the engine efficiency. Another goal is also to reduce the complexity involved in ignition timing control. 3 Materials and methods Model based predictive control (MPC) will be used as an advanced controller to control the B36:45 LNG engine. The two control inputs are the “charge air pressure” and the “global ignition timing”. The three measured input disturbances are the “IMEP”, “Charge air temperature” and “Suction air temperature”. The measured output of the system are “heat rate”, “knock”, “peak pressure”, “NOx”, “O2” and “exhaust temperature”. MPC is acting as a supervisory controller, meaning that the optimal value of these two control inputs will be further provided as set points to local PID controllers. Constraints are posed on both the inputs and the measured outputs. Heat rate is an indicative of the relationship between the power output and fuel consumption estimation. Lower value of heat rate indicates higher efficiency. Thus the MPC will be formulated as a constrained minimization problem for the heat rate. The prediction model is a data driven model in a state space form. The model is verified with a completely different data set than used for model construction. These data set are obtained from a real B36:45 engine. 4 Results The openloop simulation results show that the data driven model is capable of showing all the needed dynamics of the B36:45 engine. The MPC was able to dynamically generate optimal set points for charge air pressure and global ignition timing, while at the same time satisfying the constraints. These optimal values of the inputs increased the efficiency of the engine by lowering the heat rate. 5 Conclusions Logged data from engine tests could be successfully used to generate and verify a data driven dynamic model of the B36:45 LNG engine. It is relatively easier to regenerate or calibrate the model, and hence less complex to update the model. An advanced controller such as the MPC is able to improve the operation of the engine under various operational conditions.

ABSTRACT. Captured carbon dioxide (CO2) must be dehydrated prior to transport or storage because of possibilities for corrosion and hydrate formation. The level of accepted water content after dehydration depends on the next step for the captured CO2, and can be in the range between 5 and 500 parts per million (ppm) on a volume basis. A specification of 30 ppm is a traditional specification for pipeline transport to CO2 storage. CO2 dehydration can be performed by absorption, typically into triethylene glycol (TEG) followed by desorption or by adsorption on a solid (typically a molecular sieve) followed by desorption. It is claimed in literature that a standard TEG process can only remove water down to 30 ppm [1]. However, previous simulations have shown that glycol absorption including an extra stripping column can achieve dehydration down to less than 5 ppm [2,3]. It is of interest to perform process simulation and cost estimation to compare the two processes and evaluate how the water content specification influences the dehydration process. In this work, the process simulation program Aspen HYSYS is used to calculate material and heat balances for a TEG based absorption process and a molecular sieve adsorption process to achieve 30 ppm in the dehydrated gas. The absorption and stripping columns were modelled using a specified Murphree stage efficiency on each absorption and stripping stage. In the base case, the absorption and adsorption pressure was 30 bar, the inlet temperature was 30 °C, and the processes achieved 30 ppm in the dehydrated gas. An additional stripping column was added below the desorption column. The desorption column and the extra stripping column were simulated as one column in Aspen HYSYS with heating at an intermediate stage. Figure 1 and Figure 2 show the Aspen HYSYS flowsheets for the absorption and adsorption based processes. Both processes were cost estimated using the Aspen InPlant cost estimation tool for the equipment cost, using a detailed factor method to estimate the capital cost and typical utility cost data for heat and electricity. For the base case (with 30 ppm water in the dehydrated gas), the capital cost was calculated to 2.4 mill. EURO for the TEG unit and 5.7 mill. EURO for the molecular sieve process. The yearly operating cost was calculated to 0.1 mill. EURO for the TEG process and 0.27 mill. EURO for the molecular sieve process. The process was also calculated for dehydration down to 5 ppm. To achieve that in the TEG process, a higher absorption column is necessary. The cost of the TEG based process did not increase considerably, so the TEG absorption process was also most economical for those conditions. It is simulated reasonable process alternatives for CO2 dehydration down to water levels of 30 and 5 ppm. The simulations combined with cost estimation indicate that a TEG based process is the most economic process both for dehydration down to 30 ppm and to 5 ppm water in dehydrated gas.

ABSTRACT. A standard method of CO2 capture is absorption in monoethanolamine (MEA) followed by desorption. In this work, three configurations; standard, vapor recompression and a simple split-stream (rich split) have been simulated with an equilibrium-based model in Aspen HYSYSTM V10.0 using flue gas data from a natural gas power plant. The aim has been to calculate cost optimum process parameters and configuration and evaluate the possibility of automated cost optimization using the spreadsheet facility in Aspen HYSYSTM. This has not been evaluated for the vapor recompression or rich-split configuration in earlier work [1,2,3]. Figure 1 and 2 show simulation flowsheets including recycle and adjust blocks. The adjust and recycle blocks are used to automate the energy and material balance for a specified configuration. A spreadsheet facility in Aspen HYSYS is used for dimensioning and cost estimation of the specified process. Optimization can be performed by minimizing the total cost calculated in the spreadsheet. The cost estimation scope included installation, energy and maintenance cost of the main equipment related to flue gas cooling, absorption and regeneration. The equipment cost was obtained from Aspen In-plant Cost EstimatorTM V10.0, and an enhanced detailed factor (EDF) method [4] was used to estimate the total investment cost. Parametric studies of absorber packing height, minimum approach temperature in the main heat exchanger, flash pressure and split ratio were performed at 85 % capture efficiency. The calculated cost optimum process parameters for the standard process were 15 m packing height and 13 °C minimum approach temperature. For the vapor recompression configuration a flash pressure of 150 kPa provided the lowest total cost, with 15 m and 12 °C as optimum packing height and minimum approach temperature, respectively. The calculated optimums for the rich split configuration were 15 m packing height, 10-16 % split ratio and 9 °C minimum approach temperature. The lowest calculated capture cost of 39 Euro/ton CO2 and a reboiler energy consumption of 3.24 MJ/kg CO2 was achieved with a rich-split configuration. However, the differences in calculated capture cost for the different configurations were small. In earlier work [1,3] the vapor recompression configuration has been evaluated to be both the energy and the cost optimum configuration but they did not include the rich-split configuration. The flowsheets are simulated automatically by the help of recycle and adjust blocks when input streams and parameters (typically CO2 removal grade, pressures, temperature approach and the number of column stages) are specified. Optimization of one parameter was performed by varying the parameter in a series of simulations. For the vapor recompression case, similar optimization of the flash pressure was performed and for the rich-split the split ratio was optimized by a series of simulations. This gave more accurate optimization compared to an automated Case-study because some parameters could not be updated in the latter case. For automated cost optimization of the number of column stages, there is a limitation because the number of column stages must be (manually) specified for each new simulation. Automated calculations are dependent on stable convergence of the simulations. In principle, it should be possible to optimize the parameters in only one automated calculation with a minimization operation within the spreadsheet.

ABSTRACT. A simple, fast, and accurate process simulation based cost estimation and optimization scheme was developed in Aspen HYSYS based on a detailed factorial methodology for solvent-based CO2 absorption and desorption processes. This was implemented with the aid of the spreadsheet function in the software. The aim is to drastically reduce the time to obtain cost estimates in subsequent iterations of simulation due to parametric changes, studying new solvents/blends and process modifications. All equipment costs in a reference case are obtained from Aspen In-Plant Cost Estimator V12. The equipment cost for subsequent iterations are evaluated based on cost exponents. Equipment that are not affected by any change in the process are assigned a cost exponent of 1.0 and the others 0.65, except the absorber packing height which is 1.1. The capital cost obtained for new calculations with the Iterative Detailed Factor (IDF) model are in good agreement with all the reference cases. The IDF tool was able to accurately estimate the cost optimum minimum approach temperature based on CO2 capture cost, with an error of less than 0.2%.

ABSTRACT. The influence of different process parameters/factors on CO2 capture cost, in a standard amine based CO2 capture process was studied through process simulation and cost estimation. The most influential factor was found to be the CO2 capture efficiency. This led to investigation of routes for capturing more than 85 % of CO2. The routes are by merely increasing the solvent flow or by increasing the absorber packing height. The cost-efficient route was found to be by increasing the packing height of the absorber. This resulted in 20 % less cost compared to capturing 90 % CO2 by increasing only the solvent flow. The cost optimum absorber packing height was 12 m (12 stages). The cost optimum temperature difference in the lean/rich heat exchanger was 5℃. A case with a combination of the two cost optimum parameters achieved a 4 % decrease in capture cost compared to the base case. The results highlight the significance of performing cost optimization of CO2 capture processes.

ABSTRACT. The increasing environmental problems due to the excessive use of fossil fuels have led to implementing laws and agreements to limit global Green House Gas (GHG) emissions. Lignocellulosic biomass and biomass waste can be converted into value-added chemicals and biofuels via thermochemical or biochemical conversion. Among the different thermochemical conversion technologies, gasification is considered the cost-effective and efficient technology for lignocellulosic biomass. Gasification of biomass gives a product gas mainly consisting of syngas (CO, H₂). After gas cleaning and conditioning, the syngas obtained from biomass gasification can be used for the production of biofuels and chemicals such as methanol. Methanol is one of the important industrial chemicals that can be used directly as a fuel or can be blended into conventional fuels. A process simulation model is developed in Aspen Plus to study the conversion of syngas into methanol. A CSTR reactor was modeled with three gas-phase exothermic reactions. The product from the reactor is depressurized to separate gad from the liquid. The liquid enters into the distillation column to give CH₃OH in the distillate and the water in the bottom. Recycle stream of hydrogen is chosen to increase the overall conversion of the syngas into methanol.

ABSTRACT. This study intends to develop a mechanistic model that contributes to application of MES (microbial electrochemical synthesis) technology to increase the methane (CH4) content in the biogas produced by anerobic digestion (AD). The reactor configuration of the model consists of two reactor compartments- a continuous-flow stirred-tank reactor (CSTR) for AD process and biofilm reactor for MES process. The MES biofilm reactor is coupled to AD-CSTR with a recycle loop from it. The modelling of biogas production by AD in the CSTR follows the most used ADM-1. The modelling the MES in biofilm reactor incorporates microbially active CO2 reduction to CH4 reaction. To formulate the reduction reaction rate, the Nernst expression was incorporated as Monod-type kinetic expression, which is controlled by the electrical potential. The simulations demonstrate the basic concepts of coupling MES reactor for biogas upgrade and its limitations. The simulations show that using the MES biofilm reactor with a recycle loop from a biogas (AD) reactor increases CH4 content in the biogas. The maximum CH4 content achieved is 87 % with recycle ratios of 0.4 and 0.6 when the biofilm volume specific area is equal to 0.18 m2/m3, and with 0.6 when the specific area is equal to 0.36 m2/m3 (under the reactor condition studied). However, the conversion of CO2 to CH4 results in increased pH and consequently CH4 production decreases by ~ 40 % compared to AD-CSTR without MES. Therefore, it is essential to maintain a proper pH to prevent the inhibition on AD. The rate of the CO2 conversion to CH4 can mainly be constrained by available substrate concentration (dissolved CO2) and the local potential of the cathode and the volume specific area above 0.36 m2/m3 have minimum effects.

ABSTRACT. Aqueous pyrolysis liquid (APL) is formed from pyrolysis of lignocellulosic biomass and is considered as a possible feed for anaerobic digestion (AD). APL is known to contain many components that can have a negative impact on the AD process. In this study, APL is fed into experimental AD batch reactors and modelled as a substrate using the Anaerobic Digestion Model No. 1 (ADM1), extended by addition of the inhibitors phenol, furfural, and HMF. Simulation performed with the extended ADM1 has a better ability to predict the behavior of APL than the standard ADM1. Reducing the inhibition constants and startup concentration of active biomass during simulation of APL at high organic load resulted in improved fit with experimental results, but these inhibitors alone cannot explain the reduced methane production rate at high organic load.

ABSTRACT. We present a mathematical model of metabolism in cancer cells that is capable of describing both aerobic oxidative metabolism and anaerobic fermentation metabolism, and how cancer cells shift between these metabolic states when exposed to different substrates and different enzymatic inhibitors. The model is designed to be used in combination with experimental data gathered with an Agilent Seahorse XF metabolic analyzer. The model is parameterized in a manual tuning procedure to fit experimental data, and validated against experimental data from another setup, to which the model shows good conformity. We also investigate the structural identifiability of the model. The results indicated that the model is structurally identifiable, and that it can thus be uniquely parameterized, using the following 5 measurements: extracellular concentration of glucose, glutamine and lactate, proton production rate (a Seahorse XF analyzer measurement) and oxygen consumption rate.

ABSTRACT. A local regularity signal can be estimated from a vibration measurement with the help of the continuous wavelet transform (CWT). The resulting local regularity signal contains a lot of diagnostic information about different faults states of a machine. It is also typically a sparse signal and thus not well suited for frequency analysis using the discrete Fourier transform (DFT). In this paper, the frequency analysis of the local regularity signal is performed using the Lomb-Scargle periodogram. Another possibility is to use the methods of compressed sensing. Vibration measurements from different fault states from test rigs are utilized in validating the proposed method and comparing it with other methods. The induced fault conditions include a bearing inner ring defect and misalignment of a claw clutch. The results are compared to more traditional spectra calculated directly from the vibration measurement, such as the spectrum of the squared envelope.

ABSTRACT. Resiliency requires manufacturing system adaptability to internal and external changes, such as quick responses to customer needs, supply chain disruptions, and markets changes, while still controlling costs and quality. Sustainability requires simultaneous consideration of the economic, environmental, and social implications associated with the production and delivery of goods. Due to increasing complexity, the engineering of a production system is a knowledge-intensive process. In this paper, a summary of system adaptation methods are shown, and a holistic methodology for the assembly equipment and system modeling and evaluation is explained. The aim here is to bring resiliency and sustainability considerations into the early decision-making process. The methodology is based on estimations on system performance, using discrete event simulation run results, or other process modeling methods, and the use of Key Performance Indicators (KPI), such as Overall Equipment Efficiency (OEE), connected to cost parameters and environmental aspects analysis. Overall, it is a tool developed through multiple projects for design specification reviews and improvements, trade-off analysis, and investments justification.

ABSTRACT. X-TEAM D2D project is focused on integrating Air Traffic Management and Urban Air Mobility into an overall multimodal transport network to address the potential increase in efficiency of the overall transportation system in the future, considering the operational domain of the urban and extended urban environment up to a regional extent and passenger-centric perspective. This paper presents the analysis of the Door to Airport trajectory of business passengers until 2035. The results indicate the system's expected performance in 2035 under normal and disrupted scenarios providing insight on the expected impact of future technologies.

ABSTRACT. Electricity grid flexibility is vital for renewable energy to be used effectively. Power-to-gas technologies are investigated to connect electricity grid to gas grid and to tackle capacity challenges. Grid management expenses consist of redispatch and feed-in management. These management procedures, next to being costly, cause a significant energy loss. Proton-exchange membrane electrolyzer installations were studied to reduce these expenses and recover energy. The change in the levelized cost of hydrogen production with varying electrolyzer capacities was presented. The sensitivity of the levelized cost and net present value with respect to installation costs, maintenance costs, and electricity prices were investigated. While the electricity prices have the most significant effect on the levelized cost of hydrogen production, the net present value was affected considerably by the hydrogen selling price. Possible energy savings were calculated between 2 - 23 GWh for 2, 5, 10, 20 MW installations. The annual grid management expense savings were in the range of 0.2 - 2.3 million Euros, increasing with the increasing electrolyzer capacity.

ABSTRACT. Condensation in biogas is a problem because the condensing liquid containing CO2, water and other trace components can be very corrosive. Raw biogas typically contains 60 mol-% methane, 40 mol-% CO2, water and traces of contaminants as H2S and NH3 (ammonia). During compression up to 300 bar, condensation of water may occur. When the biogas contains NH3, it is of interest whether it has influence on the condensing (dew point) temperature. It is also of interest whether the simulations can predict how much NH3 that will end up in the condensed liquid phase. The aim of this work is to calculate the dew point of biogas containing NH3 under different conditions with varied temperature, pressure and gas composition and using different equilibrium models. Similar simulations have been performed in biogas mixtures without NH3 [1,2,3]. Traditionally, gas mixture properties of methane, CO2 and water have been calculated with standard models like Peng-Robinson (PR) and Soave-Redlich-Kwong (SRK). There is traditionally one constant binary parameter for each component pair. In the process simulation tools Aspen HYSYS and Aspen Plus, the binary parameter for water/CO2 can be made temperature dependent. Properties of mixtures of the biogas components have been studied extensively in natural gas processing and for carbon capture and storage. There have been made several models for mixtures containing ammonia with emphasis on binary systems [4]. There are a few publications about process simulation of biogas [5], but these are not considering NH3. Dry mixtures and mixtures including water have been simulated at different temperatures and pressures. Adding up to 1 mol-% NH3 to the mixtures have been simulated and dew points have been calculated with different methods and with different binary parameters. For some conditions, phase envelopes have been calculated and different models have been compared. For mixtures of methane and CO2 with up to 1 mol-% NH3 (a high value for biogas), the different models gave similar results. In earlier work, it has been shown that under normal ambient temperatures (above 0 °C), a mixture with more than 40 mol-% methane will not give any condensation. When the NH3 amount increased from 0 to 1 mol-%, the dew point temperature increased with about 3 K. In case of a more reasonable value of 0.1 mol-% NH3, the change in dew point was negligible. A phase envelope for biogas with 1 mol-% NH3 is shown in Figure 1. The phase envelope is qualitatively similar to an earlier calculated phase envelope for biogas without NH3. In biogas condensation, water will condense first. Some NH3 will dissolve in the water, and the amount of calculated NH3 dissolved in water varied significantly with the model and on the selected binary parameter for water/NH3. The solubility of NH3 in the condensate is very dependent on pH, which is influenced of the CO2 content in the biogas. Using the Peng-Robinson model with binary coefficients fitted to the NH3 solubility should give reasonable results for the NH3 solubility in water for a specific CH4/CO2 ratio of the biogas. For low concentrations of NH3 (less than 1 mol-%) the calculation of the correct dew point temperature will not be influenced by the choice of the water/NH3 binary parameter.

ABSTRACT. Amine based carbon capture is regarded as the most mature process to decrease or remove CO2 emission from coal- and gas fired power plants. The process is based upon applying an amine, especially monoethanolamine (MEA) as the most actual amine [1], to dissolve CO2 from flue gas in an absorption column shown in figure 1. The outlet solution from the bottom of absorber, rich amine, is sent to a stripper column to be regenerated and sent back to the absorber. The process can be controlled by numerous parameters. That is why various simulations and experimental studies have been conducted to improve performance of the process. Generally, process improvements can be classified into three different categories, including [2]: - Different configuration of removal process e.g. vapor recompression - Optimization of operational conditions e.g. pressure and temperature of absorber and stripper column - Switch from monoethanolamine (MEA) to other solvents or their blends. Several projects have been conducted at Telemark University College and University of South-Eastern Norway to reach an optimal removal simulation known as base case where 30% MEA solvent absorbs CO2 from flue gas [3]. MEA is one of the most important absorber liquids and the least expensive [4]. The conventional simulated process, figure 1, has been performed in a 10-stage absorber, a 6-stage desorber and 10℃ as minimum different approach temperature in the lean rich heat exchanger. The removal efficiency is 85%. The explained process could be performed with other sorts of solvents or their blends. Primary and secondary amines, like MEA, have fast reaction kinetics with CO2 but with high energy consumption to regenerate amine in the stripper. Tertiary amines, like MDEA, require less regeneration energy but they absorb CO2 slowly [5] [6]. In addition, corrosion and solvent degradation are drawbacks of MEA while for MDEA maximum loading capacity, lower corrosion and oxidative degradation than MEA are positive [6]. Piperazine (PZ) is added to increase the reaction rate. Thus, mixing amines could provide blends with less shortcomings. Other important parameters as heat of absorption, cyclic loading, CO2 lean and rich loadings are not the same for different solvents and blends. For instance, [5] experimented heat of absorption for pure amines of MEA and MDEA where MDEA solvent had lower heat of absorption and consequently lower regeneration energy. The most influential parameter for the total cost of removal plants is regeneration energy. Based on [7], this parameter accounts for up to 70% of energy demand. This study intends to simulate the effect of adding piperazine and MDEA to MEA in term of regeneration energy, cyclic capacity and CO2 loading. Carbon Dioxide removal plant process have been simulated with 3 different concentrations of (MEA+PZ) where 5 wt%, 10% wt% and 15 wt% piperazine is added to 30 wt% MEA (base case). The work proceeded with 5 different cases of MEA+MDEA blends where 5 wt%, 10 wt%, 15 wt%, 20 wt% and 25 wt% MDEA have been added and simulated to 30 wt% MEA. The results show that a blend of 30 wt% MEA + 5 wt% PZ is optimum in term of regeneration energy compared to other concentrations of MEA+PZ. Furthermore, 30 wt% MEA + 15 wt% PZ provides the lowest amount of regeneration energy among simulated cases for MEA+MDEA blends. The results are presented in figure 2 and figure 3 below. Amine blends of (30 wt% MEA+ 5% wt% PZ) and (30 wt% MEA + 15 wt% PZ) led to a decline by 4.9% and 7.5% in regeneration energy compared to base case (30 wt.% MEA) with 3.771 MJ/kg absorbed CO2. An economical study for whole simulated processes has been performed. These studies originate from mass and energy balance equations, resulting in dimensioning all equipment pieces in the plant. Aspen In-Plant Cost Estimator has been used for cost analysis. Calculated CAPEX updating material and other relevant expenses, e.g. engineering costs, direct costs and the Enhanced Detail Factor (EDF) method was applied. Besides, OPEX was calculated with the aid of extracted data from [8]. Summation of CAPEX and OPEX forms total installed costs. The applied Aspen In-Plant Cost Estimator provides data for 2018, whereas the project should be updated to 2021 so that CEPCI (Chemical Engineering Plant Cost Index) was implemented. Furthermore, both suggested blends have potential to improve the economy in a removal plant. Total amount, including OPEX and annualized CAPEX, for base case is 72.1 million Euro per year. According to economic analysis for simulated cases, both blends, (30 wt% MEA + 5 wt% PZ) and (30 wt% MEA + 15 wt% MDEA), lead to approximately 1.5% and 3.8% savings in total costs for a Carbon Dioxide removal plant which is mainly coming from reduction in required steam.

ABSTRACT. 1 Background

The Covid-19 pandemic of 2020 spurred an interest in epidemiology and prediction models for diseases. Part of this interest was due to the uncertainty the pandemic created, and a wish to understand the situation and a desire to be in “control”. Classical epidemiology models were developed in the decade following the “Spanish Flu”, and epidemiology models have developed significantly since then. During 2020, many researchers in mathematically and modeling oriented fields spent their lock-down research time in getting acquainted with epidemiology, perhaps from a different angle than classical epidemiology. These fresh approaches should be expected to provide new ideas that may be useful in handling future virus outbreaks. Epidemics mitigation can be considered as a special case of feedback control.

2 Aims

The aim of the paper is to describe epidemiological models from a control engineering perspective, where modeling for control and model-based control is essential. The main focus is on how to model epidemiological models, with a focus on population balance models and “chemical” reactions/“mass action” reactions. The aim is to give concrete illustrations on how to solve model equations.

3 Materials and methods

Model development is based on population balances with reaction rates based on mass-action kinetics, leading to deterministic ordinary differential equations for population subgroups with certain attributes. Assuming infections are governed by a Poisson distribution, the deterministic model can be developed into stochastic differential equations which describe the uncertainty of predictions. Alternatively, integer variable models based on time-to-infection (Gillespie) can be developed. Monte Carlo methods can be used to study model uncertainty, and can also be used to provide summary statistics of the model prediction. Based on available data of infection, model parameters can be estimated, e.g., using Markov Chain Monte Carlo (MCMC) methods, and by fitting models of different complexity, it is possible to assess which model best suits the data. The base concepts of epidemiology are illustrated via published data for measles infection. A brief discussion is given on how epidemiological models relate to control engineering.

ABSTRACT. 1 Background

The Covid-19 pandemic of 2020 spurred an interest in epidemiology and prediction models for diseases. Part of this interest was due to the uncertainty the pandemic created, and a wish to understand the situation and a desire to be in “control”. Classical epidemiology models were developed in the decade following the “Spanish Flu”, and epidemiology models have developed significantly since then. During 2020, many researchers in mathematically and modeling oriented fields spent their lock-down research time in getting acquainted with epidemiology, perhaps from a different angle than classical epidemiology. These fresh approaches should be expected to provide new ideas that may be useful in handling future virus outbreaks. Epidemics mitigation can be considered as a special case of feedback control.

2 Aims

The aim of the paper is to describe simple Covid-19 models from a control engineering perspective. The focus is on model fitting and mitigation policies.

3 Materials and methods

A group of models based on the SEICUR model of XXX (“SEIRUR” in the original literature) are described, with focus on fitting infection rates to experimental data, and how to account for the effect of mitigation policies from authorities for reducing the infection spread. The models under study are deterministic ordinary differential equations for a group of population attributes (S, E, I, C, U, R). A key element in the model fitting is to describe the initial, unmitigated infection growth, and how this relates to infection growth. Based on the resulting parameters, and a modification of the infection rate given by the mitigation policy of the authorities, infection spread can be predicted. Vice versa: based on certified number of infections, a combined mitigation policy and seasonal variation can be found through model fitting. Possibilities to separate the mitigation policy with seasonal/genetic variations are briefly discussed, e.g., the effect of social distancing and mask use. Other models than the SEICUR models exist, e.g., the model used by the Public Health Institute of Norway, and a brief comparison is given between the two model classes.

ABSTRACT. 1 Background

The Covid-19 pandemic of 2020 spurred an interest in epidemiology and prediction models for diseases. Part of this interest was due to the uncertainty the pandemic created, and a wish to understand the situation and a desire to be in “control”. Classical epidemiology models were developed in the decade following the “Spanish Flu”, and epidemiology models have developed significantly since then. During 2020, many researchers in mathematically and modeling oriented fields spent their lock-down research time in getting acquainted with epidemiology, perhaps from a different angle than classical epidemiology. These fresh approaches should be expected to provide new ideas that may be useful in handling future virus outbreaks. Epidemics mitigation can be considered as a special case of feedback control.

2 Aims

The aim of the paper is to describe added complexity in Covid-19 models, and the effect of vaccination.

3 Materials and methods

A group of models based on the SEICUR model of XXX (“SEIRUR” in the original literature) are described, with focus on extending the models to be (i) more general, and (ii) more useful. Possible extensions relate to network models/distributed models, and to age dependent models. Network models can also be used to study the effect of migration/tourism. A further extension is the effect of vaccination. By combining vaccination rates and mitigation policy, control engineering may suggest an “optimal” return to normalcy.

ABSTRACT. The COVID-19 pandemic is affecting around the world. There are strong differences between countries and regions. People of all ages can be infected but older people and people with pre-existing medical conditions are more vulnerable to becoming severely ill. The pandemic situation is intensively followed all over the world and there are extensive datasets including daily new confirmed COVID-19 cases and deaths for different countries. Temporal analysis of the statistical data provide early detection of changes, fluctuations, trends and severity of pandemic situations. Epidemiological models are more difficult to develop. The driving forces can any time change drastically which means that the trends can stop and even turn opposite. The temporal analysis is aimed for detecting these changes. It is an interesting question if this could be used for adapting the model parameters. The dynamic model would then be a digital twin type model where the effects of the introduced infections and mitigation are embedded. This research aims to develop intelligent data-based models where driving forces are extracted by unified intelligent temporal analysis methodologies. Parametric systems are used to adapt the solution for varying operating conditions caused by epidemics mitigation in local areas and within groups of people. Recursive updates are used in the parametric models.