Reliable Simulation Model Research Articles

Calibrating building simulation models using multi-source datasets and meta-learned Bayesian optimization

Reliable building simulation models are key to optimizing building performance and reducing greenhouse gas emissions. Informed decision making requires simulation models to be accurate, extrapolatable, and interpretable, all of which require calibrating model simulations to ground truth. Complicated building dynamics and highly uncertain exogenous disturbances make the model calibration process challenging and expensive; hence, a scalable and efficient calibration approach is needed to enable actual application. Current automatic calibration algorithms do not leverage data collected from multiple sources: for example, data obtained from previous calibration tasks on other buildings. In this paper, we employ probabilistic deep learning to meta-learn a distribution using multi-source data acquired during previous calibration. Subsequently, the meta-learned Bayesian optimizer accelerates calibration of new, unseen tasks. The few-shot (that is, requiring few model simulations) nature of the proposed algorithm is demonstrated on a Modelica library of residential buildings validated by the United States Department of Energy (USDoE). The proposed algorithm is compared against classical Bayesian optimization-based calibration, and it is shown that ANP significantly sped up the calibration procedure: the optimal model parameters are identified with 40–60% less simulations compared to the baseline.

The Investigation of Plume-Regolith Interaction and Dust Dispersal during Chang’E-5 Descent Stage

The plume-surface interaction that occurs as a result of a variable-thrust engine exhaust plume impinging on soil during landings is critical for future lunar mission design. Unique lunar environmental properties, such as low gravity, high vacuum, and the regolith layer, make this study complex and challenging. In this paper, we build a reliable simulation model, with constraints based on landing photos, to characterize the erosion properties induced by a low-thrust engine plume. We focus on the low-thrust plume-surface erosion process and erosion properties during the Chang’E-5 mission, aiming to determine the erosion difference between high- and low-thrust conditions; this is a major concern, as the erosion process for a low-thrust lunar mission is rarely studied. First, to identify the entire erosion process and its relative effect on the flat lunar surface, a one-to-one rocket nozzle simulation model is built; ground experimental results are utilized to verify the simulated inlet parameters of the vacuum plume flow field. Following that, plume flow is considered using the finite volume method, and the Roberts erosion model, based on excess shear stress, is adopted to describe plume-surface interaction properties. Finally, a Lagrangian framework using the discrete phase model is selected to investigate the dynamic properties of lunar dust particles. Results show that erosion depth, total ejected mass, and the maximum particle incline angle during the Chang’E-5 landing period are approximately 0.2 cm, 335.95 kg, and 4.16°, respectively. These results are not only useful for the Chang’E-5 lunar sample analysis, but also for future lunar mission design.

Structure of an artificial-intelligence-assisted reservoir characterization and field development protocol

Characterization of oil reservoirs for effective field development strategies is an arduous task as it requires the integration of the field data collected in various forms (for instance, seismic survey, well log and field historical pressure/production/injection data, etc.) which requires challenging workflows. The implementation of high-fidelity numerical simulators plays a vital role in quantitively identifying the sweet spots for infill drilling and optimizing the field development strategies. However, the structuring of a reliable numerical simulation model requires gridding, property upscaling and history matching, which all could be considered as laborious and time-intensive processes. Moreover, when the availability of the field data is limited, the use of high-fidelity numerical model could be prohibitively difficult, which may necessitate the development of an alternative approach to effectively characterize, integrate and design the field development plans. This paper presents a class of artificial-neural-network based expert systems trained and tested using field data collected from an oil field in North America. The proposed expert system is capable of generating artificial well logs and assessing the hydrocarbon productivity at hypothetical infill drilling locations covered by the seismic survey. Taking advantages of the robust computational efficacies of the expert systems, high-resolution heat maps indicating the productivity sweet spots can be generated to assist the decision makers in selecting the most prolific infill drilling locations when the available field data is insufficient to structure a rigorous numerical simulation model. The expert systems are trained to adapt to the data structure as exhibited by the field exploration and production data and generate realistic predictions.

FINANCIAL SUPPORT OF THE ADAPTIVE MANAGEMENT SYSTEM OF CONSTRUCTION ENTERPRISES

The article examines the impact of financial support on the management of economic entities and develops a methodical approach to assessing the level of financial support of the system of adaptive management of construction companies. It is noted that today the introduction of modern management systems into the practical activities of construction companies, capable of adapting business entities to the existing conditions of doing business, is of particular importance. It is the financial support that needs special attention in the system of adaptive management of construction companies. The main existing approaches of Ukrainian and foreign scientists to the definition of «financial security» of economic entities are considered. It is established that the financial support of construction companies has certain features related to the specifics of production and economic activities and existing management systems. It is emphasized that the financial support of construction companies significantly affects the overall management system and effective management decisions, due to the possibility of creating certain conditions to maximize net cash flow, the growth of which will increase the self-financing of the entity and reduce its dependence on external sources. The key characteristics and features of financial support of the system of adaptive management of construction enterprises are highlighted. It is concluded that the proposed methodical approach to assessing the level of financial support of the system of adaptive management of construction companies allows to build a universal reliable simulation model with a predetermined level of efficiency and the ability to make additional changes. It is proposed, in the future, to assess the level of financial security of construction companies as an element of adaptive management, which will certainly allow you to make effective management decisions on all activities of the business entity.

Accurate numerical prediction of thermo-mechanical behaviour and phase fractions in SLM components of advanced high strength steels for automotive applications

Conventional crash absorber in automotive applications, so called crash boxes are fabricated via deep drawn sheet metal resulting in significant lead times and costs. Laser Powder Bed Fusion processes, like Selective Laser Melting (SLM) offer an attractive alternative for the fabrication of crash parts while eliminating any need for costly forming dies and reducing the lead times, provided required material properties are achieved. Reliable numerical simulation model to predict the SLM build process with greater spatial resolution and accuracy is indispensable to understand the process further in order to ensure its applicability to crash structures. In this paper, an improved simulation methodology for SLM process is presented to predict the material behaviour via temperature, deformation, hardening, flow stress and phase fractions throughout the component with increased accuracy and greater resolution. To achieve desired spatial resolution, the equivalent layers are subdivided into individual tracks, which are then deposited sequentially to simulate the printing process. The material is a medium manganese (7­-8 %) transformation induced plasticity (TRIP) steel with austenite and martensite primary phases. The multiple solid-state phase transformation cycles undergone by the material are modelled in the simulation and the final phases are predicted. The results indicate improved accuracy and higher resolution in predictions for temperature, phase fractions and deformation.

Development and validation of a driving simulator for traffic control using field data

This paper presents the development and validation of a driving simulator for ramp traffic control on expressways using a traffic simulator and control (TSC). The TSC consists of two main components: car-following model (CFM), and traffic controller (TC). The CFM simulates the car-following behavior and delivers aggregated traffic parameters to the TC to derive control actions. The CFM and TC are harmonized and integrated in a close-loop control manner, where the effects of the control by the TC are fed-back as inputs for the CFM in real-time applications. Although the following behavior of individual vehicles is simulated, the aggregated outputs such as average speed and flow rate from the model are the parameters of interest. For simplicity in the model development and validation and to capture lane-changing effects, the traffic in the multi-lane expressway where the data were obtained was equivalently represented as a single-lane system. The validation of the CFM was performed at macroscopic level where aggregated outputs from the model were compared to observed data in a segment of the Pan Island Expressway of Singapore under various traffic conditions. The result shows that the simulated speed is not significantly different from the actual speed at 5% significance level, and the aggregated flow rate discrepancies fall in a small range, from 2.21% to 3.15%. This shows that the TSC model is a reliable model for traffic simulation and control applications.

Effects of 15-Day Head-Down Bed Rest on Emotional Time Perception.

Accurate time perception is clearly essential for the successful implementation of space missions. To elucidate the effect of microgravity on time perception, we used three emotional picture stimuli: neutral, fear, and disgust, in combination with a temporal bisection task to measure 16 male participants’ time perception in 15 days of –6° head-down bed rest, which is a reliable simulation model for most physiological effects of spaceflight. We found that: (1) participants showed temporal overestimation of the fear stimuli in the middle phase (day 8), suggesting that when participants’ behavioral simulations were consistent with the action implications of the emotional stimuli, they could still elicit an overestimation of time even if the subjective arousal of the emotional stimuli was not high. (2) Participants’ temporal sensitivity tends to get worse in the bed rest phase (days 8 and 15) and better in the post-bed rest phase, especially for neutral and fear stimuli, suggesting that multiple presentations of short-term emotional stimuli may also lead to a lack of affective effects. This reduced the pacemaker rate and affected temporal perceptual sensitivity. Also, this may be related to changes in physiological factors in participants in the bed rest state, such as reduced vagal excitability. These results provide new evidence to support the theory of embodied cognition in the context of time perception in head-down bed rest and suggest important perspectives for future perception science research in special environments such as microgravity.

Deep Learning-Based Predictive Framework for Groundwater Level Forecast in Arid Irrigated Areas

An accurate groundwater level (GWL) forecast at multi timescales is vital for agricultural management and water resource scheduling in arid irrigated areas such as the Hexi Corridor, China. However, the forecast of GWL in these areas remains a challenging task owing to the deficient hydrogeological data and the highly nonlinear, non-stationary and complex groundwater system. The development of reliable groundwater level simulation models is necessary and profound. In this study, a novel ensemble deep learning GWL predictive framework integrating data pro-processing, feature selection, deep learning and uncertainty analysis was constructed. Under this framework, a hybrid model equipped with currently the most effective algorithms, including the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) for data decomposition, the genetic algorithm (GA) for feature selection, the deep belief network (DBN) model, and the quantile regression (QR) for uncertainty evaluation, denoted as CEEMDAN-GA-DBN, was proposed for the 1-, 2-, and 3-month ahead GWL forecast at three GWL observation wells in the Jiuquan basin, northwest China. The capability of the CEEMDAN-GA-DBN model was compared with the hybrid CEEMDAN-DBN and the standalone DBN model in terms of the performance metrics including R, MAE, RMSE, NSE, RSR, AIC and the Legates and McCabe’s Index as well as the uncertainty criterion including MPI and PICP. The results demonstrated the higher degree of accuracy and better performance of the objective CEEMDAN-GA-DBN model than the CEEMDAN-DBN and DBN models at all lead times and all the wells. Overall, the CEEMDAN-GA-DBN reduced the RMSE of the CEEMDAN-DBN and DBN models in the testing period by about 9.16 and 17.63%, while it improved their NSE by about 6.38 and 15.32%, respectively. The uncertainty analysis results also affirmed the slightly better reliability of the CEEMDAN-GA-DBN method than the CEEMDAN-DBN and DBN models at the 1-, 2- and 3-month forecast horizons. The derived results proved the ability of the proposed ensemble deep learning model in multi time steps ahead of GWL forecasting, and thus, can be used as an effective tool for GWL forecasting in arid irrigated areas.

Physics-Aware Deep-Learning-Based Proxy Reservoir Simulation Model Equipped With State and Well Output Prediction

Data-driven methods have been revolutionizing the way physicists and engineers handle complex and challenging problems even when the physics is not fully understood. However, these models very often lack interpretability. Physics-aware machine learning (ML) techniques have been used to endow proxy models with features closely related to the ones encountered in nature; examples span from material balance to conservation laws. In this study, we proposed a hybrid-based approach that incorporates physical constraints (physics-based) and yet is driven by input/output data (data-driven), leading to fast, reliable, and interpretable reservoir simulation models. To this end, we built on a recently developed deep learning–based reduced-order modeling framework by adding a new step related to information on the input–output behavior (e.g., well rates) of the reservoir and not only the states (e.g., pressure and saturation) matching. A deep-neural network (DNN) architecture is used to predict the state variables evolution after training an autoencoder coupled with a control system approach (Embed to Control—E2C) along with the addition of some physical components (loss functions) to the neural network training procedure. Here, we extend this idea by adding the simulation model output, for example, well bottom-hole pressure and well flow rates, as data to be used in the training procedure. Additionally, we introduce a new architecture to the E2C transition model by adding a new neural network component to handle the connections between state variables and model outputs. By doing this, it is possible to estimate the evolution in time of both the state and output variables simultaneously. Such a non-intrusive data-driven method does not need to have access to the reservoir simulation internal structure, so it can be easily applied to commercial reservoir simulators. The proposed method is applied to an oil–water model with heterogeneous permeability, including four injectors and five producer wells. We used 300 sampled well control sets to train the autoencoder and another set to validate the obtained autoencoder parameters. We show our proxy’s accuracy and robustness by running two different neural network architectures (propositions 2 and 3), and we compare our results with the original E2C framework developed for reservoir simulation.

Experimental Methods for Measuring the Viscous Friction Coefficient in Hydraulic Spool Valves

In hydraulic components, nonlinearities are responsible for critical behaviors that make it difficult to realize a reliable mathematical model for numerical simulation. With particular reference to hydraulic spool valves, the viscous friction coefficient between the sliding and the fixed body is an unknown parameter that is normally set a posteriori in order to obtain a good agreement with the experimental data. In this paper, two different methodologies to characterize experimentally the viscous friction coefficient in a hydraulic component with spool are presented. The two approaches are significantly different and are both based on experimental tests; they were developed in two distinct laboratories in different periods of time and applied to the same flow compensator of a pump displacement control. One of the procedures was carried out at the Fluid Power Research Laboratory of the Politecnico di Torino, while the other approach was developed at the University of Parma. Both the proposed methods reached similar outcomes; moreover, neither method requires the installation of a spool displacement transducer that can significantly affect the results.

Three-dimensional finite element simulation of high speed milling of titanium alloy Ti6Al4V

In this paper, ABAQUS finite element simulation software was used to establish a more reliable three-dimensional milling titanium alloy Ti6Al4V finite element simulation model. This model sets up important links such as the material constitutive model, the tool-chip contact friction model, and the chip separation criterion. And the accuracy of the model is verified by experiments. Based on this model, orthogonal experiment was used to explore the effects of axial cutting depth, tool speed and feed speed on the milling force. And the morphology of the chip and the surface morphology of the machined workpiece were investigated. The results show that the influence of cutting parameters on milling force from high to low is as follows: axial cutting depth > tool speed > feed rate. The change of cutting parameters has different influence on milling force in all directions, and has the largest influence on milling force in X direction.

Design and parameters optimization of root cutting tool based on garlic numerical simulation model

AbstractTo obtain the best tool parameters for garlic root cutting to reduce the root cutting force and improve efficiency, a linkage occlusion‐cutting mechanism was designed for a garlic root cutting force test. The physical and mechanical properties of garlic meat and garlic skin were measured and calculated by using a universal material testing machine. A plastic follow‐up model (MAT‐3) in the LS‐DYNA material library was used as a simulated garlic material model. A three‐dimensional model of the garlic and the tool was established and imported into ANSYS/LS‐DYNA to establish a garlic root cutting simulation model. The model was used for numerical calculation and analysis. The simulation parameters were optimized according to the actual cutting force change curve, and a reliable simulation model was obtained. The error between simulated result and actual measurement was 3.5%. Based on this reliable model, the effects of tool parameters on the root cutting force were explored. The optimal blade edge angle was 11°, the optimal blade back angle was 13°, and the cutting root force was reduced by 14.94%. The reliable model and optimal tool parameters provide a reference for the design and development of garlic root cutting equipment.Practical ApplicationsGarlic is popular worldwide because of its high nutritional value and unique flavor. A reliable numerical simulation model for garlic will encourage researchers to design garlic‐processing equipment. The optimal tool parameters also provide a reference for garlic root cutting equipment.

Reducing Power Consumption of Active-Matrix Mini-LED Backlit LCDs by Driving Circuit

An active-matrix (AM) driving circuit using a low-temperature poly-silicon thin-film transistor (LTPS TFT) is developed to reduce the power consumption of mini-light-emitting diode (LED) backlit liquid-crystal displays (LCDs). By lowering the drain-source voltage ( V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DS</sub> ) of the switching TFT on the driving current path, the range of required voltage across the proposed circuit can be reduced, decreasing the power consumption of the mini-LED backlight. The proposed circuit also can compensate for V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">TH</sub> variation in the LTPS TFT and the I- R rise of V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">SS</sub> to provide a uniform driving current to flow through the mini-LED. By measuring the I- V curves of a fabricated LTPS TFT, we build a reliable simulation model and compare the proposed circuit to that of the state-of-the-art. Simulated results indicate that the proposed circuit can improve the power consumption by 16.67% and reduce the current error rates below 8% when the V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">TH</sub> of TFT varies by ±0.3 V and V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">SS</sub> rises +1 V. Measured results further confirm that the V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DS</sub> of the switching TFT is reduced by 2.208 V compared to a 6T2C compensating circuit. Therefore, widespread applications of the proposed circuit for AM mini-LED backlit LCDs to produce high-dynamic-range images are foreseeable.

Development of Simulation for Predicting Maintenance Costs for Educational Facilities

In Korea, educational facilities are managed after issues are detected. Furthermore, due to imprecisely estimated maintenance costs, there are problems in managing educational facilities such as excessive budgets, etc. it is impossible to accurately estimate costs for various cases such as repairs and damages. Thus, the overall maintenance costs were analyzed based on probability of cases that may occur. This study proposes a system that predicts maintenance subjects and maintenance costs using probability values and that predicts the total maintenance cost through simulations. This study used simulations to accurately predict maintenance costs in various unexpected situations. The study was carried out by classifying into three categories according to each feature. The features were described according to the type of each item and simulation of the categorized items were produced separately to calculate the expected price. In the case of ‘annual fixed expenses’, the total of the maintenance cost tended to increase with a secondary curve where the slope gradually grows with time. In the case of ‘irregular expenses due to repairs/changes caused by minor failures’, maintenance cost had a certain cycle according to a failure rate according to the life of each item and an inflation rate of 1.9% was applied to show a tendency for occurrence to grow. In the case of ‘Replacement/repair costs according to the scheduled period for the entire educational facility’, the simulation results showed that each item had a constant cycle depending on the replacement period and that facility maintenance costs did occur. This study presumed data for simulations, but did not verify it. By supplementing such limitations, it will be possible to produce a reliable simulation model. The expected effect of reducing budget waste by predicting the exact cost of educational facilities through simulation.

Reliability Evaluation Method of Torpedo in Service Based on Conditional Reliability

In order to effectively evaluate the reliability index of in-service torpedo, combined with the actual use of torpedo, the reliability evaluation method of in-service torpedo based on conditional reliability is proposed. On the basis of determining the failure distribution under the single state of components, the variation of the reliability distribution of the components after several states are studied. Based on the fault tree structure function, the reliability simulation model of torpedo is established, and the Monte Carlo method is used to establish the simulation solution process, so as to obtain the reliability parameters of torpedo. This method is simple and easy to implement, and has important guiding significance for evaluating the reliability of in-service torpedoes.

Calibration and validation of the dynamic response of two slab track models using data from a full-scale test rig

For the development of accurate and reliable simulation models, the procedure of calibration and validation against measurement data is essential. In this paper, a finite element model and a waveguide finite element model of a slab track are calibrated and validated against hammer impact measurement data from a full-scale test rig. The finite element model is three-dimensional, where the rails are modelled as Rayleigh–Timoshenko beams and the concrete slab and support layer are modelled using linear shell elements. In the waveguide finite element model, a constant track cross-section described by two-dimensional finite elements is assumed, and the vibration in the direction perpendicular to the cross-section is described by propagating waves that are decaying exponentially. Measured frequency response functions (FRFs) are compared with the corresponding calculated FRFs from the two modelling approaches. The calibration is conducted in two steps using (i) a parameter study and (ii) a genetic algorithm. For multiple excitation positions and sensor locations, both calibrated models capture the trend of the Single-Input Multiple-Output measurements with rather small deviations compared to the overall dynamic range. This implies that both models can successfully represent the dynamic response of the test rig and can be considered as validated.

Novel method for the parameterization of a reliable equivalent circuit model for the precise simulation of a battery cell's electric behavior

This article presents a novel approach to extract the equivalent circuit model parameter from a pulse test. The presented results corroborate that this novel heuristic methodology can very accurately fit the voltage response of the battery, while drastically reducing the dependence of the parametrization on set boundary conditions and initial guess values. The method's robustness is verified by fitting the voltage response of a mock battery model with pre-defined parameters. It could be shown that the suggested method is able to precisely extract the set parameters from the simulated voltage. Furthermore, the model values obtained from the test of a lithium ion battery cell correlate with physical parameter and other investigation methods like the calculation of the distribution of relaxation times (DRT). As a result, an accurate equivalent circuit model of the cell was created. The restriction as to universal validity of the empirical equivalent circuit model was significantly reduced by limiting the voltage of the RC element with highest time constant. Finally, a robust battery cell model with only a minor root-mean-square error of 1.30% was obtained. ResultA new method for the precise parameterization of an equivalent circuit model from pulse tests is presented. The robustness of such equivalent circuit models is increased by limiting the voltage of the RC element with the highest time constant. Overall, a precise, reliable model of a lithium-ion battery cell is created.

Development of advanced explosion safety technology for hydrogen systems

A team of researchers from The University of Tokyo, Yamagata University and Hiroshima University, are working to understand phenomena related to large-scale hydrogen gas explosions. The team's research aims to develop a database of gas explosion behaviour through hydrogen gas explosion experiments. The hope is that this will deepen academic understanding, enable the building of a reliable simulation model and, ultimately, develop a practical and advanced explosion safety technology that can be used when designing, developing and implementing hydrogen systems. The findings will lead to improved safety and security for industries implementing hydrogen systems.

Augmentation of Low Voltage Ride through (LVRT) Capability using ANN based electric spring for a Grid-tied Wind Energy Conversion System

Electric Spring (ES) is an innovative approach for distributed voltage control (DVC) whereas at the same time accomplishing nominal demand-side management (DSM) technique for continuously varying renewable energy sources. In order to demonstrate the efficiency of ES when deployed in enormous quantities across the grid-connected power distribution system, it is important to build easy and reliable simulation models of ES that can be integrated into large-scale simulation studies of the power system. The ES study gap identification method defines the ES dynamic simulation methodology that is appropriate for the WECS grid-connected voltage and frequency control studies. The integration of WECS to the electric grid leads to many serious power quality problems. Low voltage ride through (LVRT) capability improvement is one of the important constraints in grid connected WECS. This articleexamines the enhancement of LVRT capability of Doubly Fed Induction Generator (DFIG) based grid connected WECS by usingES. The effectiveness of proposed Artificial Neural Network (ANN) based Electric spring controller is compared with existing PI controller using MATLAB/Simulink software. It is evident that the proposed controller nurtures the trustworthiness of WECS when integrated with existing electric power grid system.

OPTIMIZATION BUILDING PERFORMANCE IN EARLY DESIGN STAGE USING INTEGRATED DYNAMIC MODEL

Considering the magnitude of energy loss in building, development of energy saving methods appears to be essential. Daylight plays a significant role in designing energy efficient buildings and improving visual comfort for the occupants. Many daylight analysis methods have been developed in this area. Most of these methods focus on opening maximization. These methods unfortunately might reduce comfort since it causes direct solar glare. There is a need for a reliable lighting simulation model to control the lighting strategy in early stage design. This study proposes a strategy for visualizing daylight analysis of buildings by using Integrated Dynamic Model (IDM). IDM is a combination of design tools used during the conceptual phase for holistic classroom that considers the building’s energy usage, daylight distribution, and thermal indoor environment. The optimization focus is related maximize the performance of the building envelope design. The purpose of this paper are; firstly, providing a new strategy for visualizing the predicting daylight while respecting architectural integrity. The second purpose is to facilitate the designer for choosing window and envelope design alternatives during early stages. The third is to maximize the positive impacts of daylight. Lastly, hopefully IDM could present a simplified simulation and analyze method with the timely, accurate and efficient process.