Search

This article presents a conveyor belt fire classification model that allows for the determination of the most effective firefighting strategy. In addition, the effect of belt design parameters on the fire classification was determined. A methodology that involves the use of numerical simulations and artificial neural networks was implemented. An approach previously proposed for modeling fires over conveyor belts was used. With the objective of obtaining some required modeling input parameter and verifying the capacity of this approach to get realistic results, computational fluid dynamics model calibration and validation were carried out using experimental test results available in the literature. Results indicated that scenarios with belt positions closer to the mine roof and greater tunnel heights require a higher longitudinal air velocity to be attacked directly. Furthermore, the belt fire classification model provided by the artificial neural network had an accuracy around 95% when test scenarios were classified.

Firebrands are known to cause ignition of structures far from the primary fire front, resulting in significant damage to structures before firefighting can be attempted. To make structures more resilient to firebrand ignition, a better understanding of the heat transfer from firebrands to surfaces is needed. This paper provides a statistical assessment of different factors expected to have an impact on the heat flux from firebrand piles to a flat surface. The factors included in the study were wood moisture content, wood type (hardwood or softwood), wood density, wood state (live, dead, or artificial), wind speed, pile mass, firebrand diameter, and firebrand length. Using design of experiments, test matrices were developed that permitted a statistical analysis to be performed on the data. This statistical analysis was used to quantify which factors had a statistically significant impact on the heat flux from the pile as well as ranking the importance of the different factors. Artificial firebrands were found to have statistically higher heat fluxes compared with natural firebrands. Other factors that had a statistically significant impact on the heat flux were wind speed, firebrand length, and firebrand length-diameter interaction. Firebrand aspect ratio (related to the firebrand length-diameter interaction) is directly related to the pile porosity, which is a measure of the volume of air in the pile. Increasing the aspect ratio (which increases the pile porosity) results in higher heat fluxes across a larger region of the pile and was found to be an important factor. Firebrand diameter and pile mass were found to affect the burning duration but not as significantly as other parameters. The number of firebrands in the pile was also observed to potentially affect the heat flux, with a critical number required to reach the highest heat flux for a given firebrand geometry. National Institute of Standards and Technology (NIST)National Institute of Standards & Technology (NIST) - USA [70NANB19H052] Published version This research was funded by the National Institute of Standards and Technology (NIST) under contractNISTGrant No. 70NANB19H052.

The computational cost of predicting wildland fire spread across large, diverse landscapes is significant using current models, which limits the ability to use simulations to develop mitigation strategies or perform forecasting. This paper presents a machine learning approach to estimate the time-resolved spatial evolution of a wildland fire front using a deep convolutional inverse graphics network (DCIGN). The DCIGN was trained and tested for wildland fire spread across simple homogeneous landscapes as well as heterogeneous landscapes having complex terrain. Data sets for training, validation, and testing were created using computational models. The model for homogeneous landscapes was based on a rate of spread from the model of Rothermel, while heterogeneous spread was modeled using FARSITE. Over 10,000 model predictions were made to determine burn maps in 6 h increments up to 24 h after ignition. Overall the predicted burn maps from the DCIGN-based approach agreed with simulation results, with mean precision, sensitivity, F-measure, and Chan–Vese similarity of 0.97, 0.92, 0.93, and 0.93, respectively. Noise in the input parameters was found to not significantly impact the DCIGN-based predictions. The computational cost of the method was found to be significantly better than the computational model for heterogeneous spatial conditions where a reduction in simulation time of $$10^{2}{-}10^{5}$$ was observed. In addition, the DCIGN-based approach was shown to be capable of predicting burn maps further in the future by recursively using previous predictions as inputs to the DCIGN. The machine learning DCIGN approach was able to provide fire spread predictions at a computational cost three orders of magnitude less than current models.

Composite metal foam (CMF) is a novel lightweight metal matrix composite material with lightweight, high strength to density ratio and high energy absorption capabilities. The material can be made out of many different metals, alloys, and combinations, e.g. aluminum, steel, titanium, etc. For example, it can be made 100% out of steel, but, due to its porosities, it will weigh as light as aluminum. CMF is made of closely packed metallic hollow spheres with a metallic matrix that fills the empty spaces in between spheres. In every combination of the spheres and matrix materials, the final product weight will be ~30–35% of the weight of the parent material; the rest would be the air trapped inside its porosities. In this study, a scaled-down version of the torch fire experiments specified in 49 Code of Federal Regulation (CFR) Part 179, Appendix B was developed to provide initial data on evaluating the thermal protection performance of steel-steel composite metal foam (S-S CMF) in the torch fire conditions. S-S CMF panels of 30 × 30 cm dimensions are manufactured and tested to evaluate their survivability when exposed to a 30-minute torch fire condition of high velocity jet fire with a gas temperature of 1204°C in accordance with 49 CFR Part 179. Testing was performed to characterize the jet burner gas temperature and velocity flow field, and a calibration fire test was conducted using steel only as required by the test specification. The assembly was tested in duplicate in two consecutive simulated torch fire exposures as specified in 49 CFR Part 179, Appendix B. Based on the experimental results, a 15 mm thick steel-steel composite metal foam tested as novel insulation system met the acceptance criteria for the simulated torch fire testing and is expected to pass when tested at a full size of 122 × 122 cm dimensions. The main factor for fire resistance and thermal protection performance of S-S CMF is attributed to the large air content in the material.

Firebrands are an important mechanism of fire spread and one of the primary ways in which wildland fires ignite structures. Inverse heat transfer using thin steel plates has been shown to be an effective method for measuring heat transfer distributions from firebrands. To fully understand the dynamic process of heat transfer from firebrands to surfaces; however, it is necessary to view the underside of the firebrands, which is not possible through a steel plate. This work develops a method of inverse heat transfer using a visually transparent quartz plate and a long-wave (7.5–14.0 μm) infrared camera to facilitate visual access to the firebrands from all angles. The heat flux measurements using a quartz plate were compared with heat flux measurements using a steel plate and finite element heat transfer simulations for radiation-dominant tests using heater panels. Additionally, heat transfer measurements using cuboidal oak firebrands were conducted using both the quartz and steel plates. A corrective factor was developed based on the ratio of the effective emissivity of the quartz and stainless-steel plates at typical firebrand temperatures. The measured heat fluxes were within 1–6% after correcting for radiant energy transmitted through the quartz which was absorbed by the stainless-steel plate.

Understanding the parameters that affect firebrand burning conditions is needed to quantify and model heat transfer from firebrands to combustible surfaces. In this research, an experimental and analytical effort was conducted to determine the variable relationships that control firebrand burning. A series of experiments were performed to quantify the mass loss rate, temperature, and char diameter change with time for single and arrays of cylindrical firebrands. An analytical model was developed to predict the time dependent burning of firebrands including ash accumulation in forced flow conditions. Six different methods for predicting char oxidation were included in the model to identify the best approach for predicting firebrand burning. Based on the simulation results, the model with char oxidation determined using the heat and mass transfer Reynolds analogy provided the best results with predicted temperatures, char diameter, and mass loss rates within 5%, 4%, and 29% of the single firebrand test data, respectively. Higher differences were predicted with arrays of firebrands, which was attributed to the complex flow field that develops around the firebrands. Analysis of the analytical equations was used to identify the variable relationships affecting firebrand temperature, mass loss rate, char diameter, and burning duration.

Elevated temperature exposure of 5XXX alloys causes sensitization, which leads to changes in room temperature mechanical properties and may lead to intergranular corrosion (IGC) susceptibility. The influence of sensitization and corrosion damage on the creep response of 5083-H116 at various temperatures (representative of fire damage) from 200 °C to 400 °C has been examined in this experimental study. Sensitization had a negligible effect on creep properties, while corrosion damage led to a reduction in creep rupture time and steady state creep rates. After accounting for sensitization and corrosion damage, creep rupture can be predicted based upon secondary creep/steady state creep rates.

Predicting the temperature of an exposed object or even a person is one of the most common tasks of fire safety engineering. However, the nonlinear nature of heat transfer and the challenge of chan ...

The simulation of large complex dynamical systems such as a fire in road tunnels is necessary but costly. Therefore, there is a crucial need to design efficient models. Coupling of computational fluid dynamics (CFD) models and 1D network modeling simulations of a fire event, a multiscale method, can be a useful tool to increase the computational efficiency while the accuracy of simulations is maintained. The boundary between a CFD model (near field) and a 1D model (far field) plays a key role in the accuracy of simulations of large systems. The research presented in this paper develops a novel methodology to select the interface boundary between the 3D CFD model and a 1D model in the multiscale simulation of vehicle fire events in a tunnel. The development of the methodology is based on the physics of the fluid structure, turbulent kinetic energy of the dynamical system, and the vortex dynamics. The methodology was applied to a tunnel with 73.73 m(2) cross section and 960 m in length. Three different vehicle fire scenarios were investigated based on two different heat reslease rates (10 MW and 30 MW) and two different inlet velocities (1.5 m/s and 5 m/s). all parameters upstream and downstream of the fire source in all scenarios were investigated at t = 900 s. The effect of changes in heat release rate (HRR) and air velocity on the selection of an interface boundary was investigated. The ratio between maximum longitudinal and transversal velocities was within a range of 10 to 20 in the quasi-1D region downstream of the fire source. The selected downstream interface boundary was 12D(h) m downstream of the fire for the simulations. The upstream interface boundary was selected at 0.5 D-h m upstream the tip of the object when the velocity was greater than equal to the V-c. In the simulations with backlayering (V < V-c), the interface boundary was selected 10 m further from the tip of the backlayering (1.2 D-h). An indirect coupling strategy was utilized to couple CFD models to 1D models at the selected interface boundary; then, the coupled models results were compared to the full CFD model results. The calculated error between CFD and coupled models for mean temperature and velocity at different cross sections were calculated at less than 5%. The findings were used to recommend a modification to the selection of interface boundary in multiscale fire simulations in the road tunnels and more complex geometries such as mines. National Institute for Occupational Safety and Health (NIOSH) [200-2014-59669] This research was developed under Contract No. 200-2014-59669, awarded by the National Institute for Occupational Safety and Health (NIOSH). The findings and conclusions in this report are those of the authors and do not reflect the official policies of the Department of Health and Human Services; nor does mention of trade names, commercial practices, or organizations imply endorsement by the U.S. Government.

A method of sensor fusion was developed to combine long-wavelength infrared (LWIR) stereo vision and a spinning LIDAR for improved rangefinding in smoke-obscured environments. This method allows rangefinding in clear and smoke conditions, relying on LIDAR’s high accuracy in clear conditions and the perception ability of LWIR cameras in smoke. Sensor data were combined using evidential (Dempster–Shafer) theory in a 3D multi-resolution voxel domain for occupied and free space states. A heuristic method was produced for separating significantly attenuated and low-attenuation LIDAR returns using return intensity and distance. A sensor model was developed to apply free space state information from LIDAR high-attenuation returns. Sensor models were developed for applying occupied and free space state information from LIDAR low-attenuation returns and from LWIR stereo vision points. The fusion method was evaluated in two fire environments: a room-hallway scenario with a range of clear to dense-smoke conditions and a shipboard fire scenario. Room-hallway tests were evaluated by assessing performance against baseline rangefinding. For the occupied state, the fusion method and LIDAR are within typically 5% to 10% for clear conditions, and the fusion method is more accurate than the LIDAR by typically 5% to 10% for smoke conditions, with LIDAR providing no data in the densest smoke. For the free space state, the fusion method outperformed the LIDAR in smoke conditions by as much as 40% and was typically within 5% of the LIDAR in clear conditions.

Inverse heat transfer analysis (IHT) was used to measure the full-field heat fluxes on a small scale (0.9 m×0.9 m×0.9 m) stainless steel SS304 compartment exposed to a 100 kW diffusion flame. The measured heat fluxes were then used in a thermo-mechanical finite element model in Abaqus to predict the response of an aluminum 6061-T6 compartment to the same exposure. Coupled measurements of deflection and temperature using Thermographic Digital Image Correlation (TDIC) were obtained of an aluminum compartment tested until collapse. Two convective heat transfer coefficients, h =35 W/m2-K and h =10 W/m2-K were examined for the thermal model using the experimentally measured heat fluxes. Predictions of the thermal and structural response of the same compartment were generated by coupling Fire Dynamics Simulator (FDS) and Abaqus using the two values for h, h =35 W/m2-K and h from convection correlations. Predictions of deflection and temperature using heat fluxes from IHT and FDS with h=35 W/m2-K agreed with experimental measurements along the back wall. The temperature predictions from the IHT-Abaqus model were independent of h, whereas the temperature predictions from the FDS-Abaqus model were dependent on h.

© 2017 Wiley Periodicals, Inc. The Electric Series Compliant Humanoid for Emergency Response (ESCHER) platform represents the culmination of four years of development at Virginia Tech to produce a full-sized force-controlled humanoid robot capable of operating in unstructured environments. ESCHER's locomotion capability was demonstrated at the DARPA Robotics Challenge (DRC) Finals when it successfully navigated the 61 m loose dirt course. Team VALOR, a Track A team, developed ESCHER leveraging and improving upon bipedal humanoid technologies implemented in previous research efforts, specifically for traversing uneven terrain and sustained untethered operation. This paper presents the hardware platform, software, and control systems developed to field ESCHER at the DRC Finals. ESCHER's unique features include custom linear series elastic actuators in both single and dual actuator configurations and a whole-body control framework supporting compliant locomotion across variable and shifting terrain. A high-level software system designed using the robot operating system integrated various open-source packages and interfaced with the existing whole-body motion controller. The paper discusses a detailed analysis of challenges encountered during the competition, along with lessons learned that are critical for transitioning research contributions to a fielded robot. Empirical data collected before, during, and after the DRC Finals validate ESCHER's performance in fielded environments.

As specified in NFPA 130, rail car floor assemblies are currently required to undergo a large-scale furnace test to demonstrate their fire resistance, including structural integrity and limited heat transmission. This results in an expensive fire resistance compliance process mostly due to the physical size requirement of test article and the size of the furnace required to test it. Although there is an interest from railcar manufacturers to reduce the physical size of the rail car floor assembly test article, there is no demonstrated approach to justify the use of a smaller test article. This work used finite element modelling to develop an approach which can be used to reduce the size of test article for fire resistance tests. The results of the analysis indicate that the support boundary conditions of the test article should be modified so the test article better represents the floor end-use response. In addition, the test article size can be reduced to a one-third of the current size.

Construction materials exposed to fire conditions may decompose resulting in a loss of strength as well as contributing to the growth and size of the fire. Predicting the behavior of timber and other construction materials during a fire exposure requires accurate thermal modeling of the material considering both the effects of elevated temperature and material decomposition state (virgin to char) on the thermal and physical properties. This paper evaluates the use of different techniques to measure the porosity, permeability and thermal diffusivity of wood construction materials at different levels of decomposition due to a high temperature exposure. The porosity and thermal diffusivity values of wood exhibit similar changes with decomposition level. It was found that virgin and char material properties can be used in conjunction with thermogravimetric data to estimate material properties at intermediate decomposition levels. The findings can be used to reduce the number of experiments required to characterize other materials. Permeability was observed to be dependent on the structure and composition of the material. Scanning electron microscopy images revealed that decomposition increased the space between the otherwise compact-arranged elements (grain, particles), resulting in an increase in permeability. Consequently, permeability should be measured at different decomposition levels and direction to quantify the material changes due to a high temperature exposure.

Aluminum alloys undergo permanent material property degradation that results in a reduced post-fire capacity of a structure. This degradation occurs at temperatures as low as 200 °C and is a function of both exposure temperature and duration. Material models have been previously developed that account for both temperature and duration but previous modeling efforts for aluminum structures have only considered maximum temperature. This research utilized an Arrhenius kinetics model to predict the material property degradation for AA6061 caused by the thermal exposure. The model was calibrated using post-fire yield strength data at different linear heating rates. The method was validated by modeling a set of small-scale experiment of an aluminum beam using thermal and mechanical finite element analyses. The beams were subjected to nominal 50 kW/m2 and 70 kW/m2 radiant heating from underneath for up to 20 min, water quenched, and loaded in 4-point bending. Beam temperatures were predicted within 25 °C during the transient response and 10 °C during the steady-state response. The predicted peak bending loads for all tested exposure levels and durations were within 8% of the experimentally measured values. Continued degradation of beam strength observed in the experiments during thermal steady-state was also captured by the kinetics-based model.

Firebrands are known to cause spot fires and structure ignition far from the fire front, but there is a limited understanding of the heat transfer from firebrands to surfaces. In this work, high resolution heat flux distributions were measured for single firebrands with different geometries using IR thermography and inverse heat transfer analysis. Localized heat fluxes from a single firebrand were measured to be 30–105 kW/m2, which is 2–6 times higher than previous work with heat flux gauges and energy balance methods that spatially average the heat transfer from the firebrand. Firebrand geometry, wind speed, and wind speed orientation relative to the firebrand affect the heat flux magnitude and duration of the exposure.

A computer vision-based autonomous fire suppression system with real-time feedback of fire size and spray direction is presented in this paper. The system has been developed for use in a firefighting robot for close-range, localized fire suppression tasks in enclosed environments. A probabilistic water classification method was developed for segmenting water spray in a pair of IR cameras. Stereo processing was performed to localize points along the spray path for use in yaw and pitch angle estimation. A Golden Section Search with linear least squares optimization was used to determine the optimal pitch angle of the spray position at each sampling time. Kalman filtering was used to remove noise from the angle measurements and obtain a better estimate of the current nozzle orientation. A decision tree was used to determine the correct nozzle positioning mode using image feedback to suppress the fire and accounts for errors in direction, fire size during suppression, and when to adjust the nozzle based on IR feedback. Through implementation of a PI controller, the system is able to correct for unknown disturbances causing erroneous targeting of a localized fire. Experiments are presented with the initial nozzle angled correctly and with forced offsets in the system to set the initial spray position incorrectly in order for the system to correct. Suppression times ranged from 7.2 s to 16.3 s with a standard deviation of 3.9 s and average time of 11.2 s. A total of 12 tests demonstrated performance of the system given a forced offset to the initial nozzle orientation resulting in an error between the spray location and the fire target. Suppression times ranged from 8.1 s to 27.9 s with a mean of 16.9 s and standard deviation of 6.2 s. The proposed system can be implemented on a robotic firefighting platform to autonomously detect a fire, choose a proper manipulation goal and suppress full scale fires given disturbances causing erroneous targeting.

An autonomous fire suppression system was developed for localized fire suppression in high and low visibility environments. The system contains a multispectral sensor suite, including UV sensors and infrared stereovision, to detect and target a fire for suppression. The UV sensor provides an alert to the system to begin fire detection. IR imagery is used to segment fire from the field of view and target the base of the fire and IR stereovision to determine the 3D coordinates of the fire. IR tracking provides continuously updated information on the size and intensity of the fire before and during suppression and alerts the system when to cease suppression activity. Visual servoing is used to correctly position a nozzle based on feedback of changes in the fire location and size. The autonomous system was used to suppress wood crib fires (40 kW to 50 kW) in high and low visibility environments and at varying distances (2.8 m to 5.5 m) and elevations (0.4 m to 1.3 m). The suppression time in clear conditions was 3.72 s ± 1.51 s and 4.49 s ± 1.62 s in low visibility conditions. To simulate wind effects and inaccurate initial target coordinates, forced offsets were input to the system to show effectiveness of the feedback control algorithm when an initial estimate of spray trajectory does not accurately spray the center base of the fire. System performance with a forced offset resulted in suppression times of 4.11 s ± 0.84 s.

Locating a fire inside of a structure that is not in the direct field of view of the robot has been researched for intelligent firefighting robots. By classifying fire, smoke, and their thermal reflections, firefighting robots can assess local conditions, decide a proper heading, and autonomously navigate toward a fire. Long-wavelength infrared camera images were used to capture the scene due to the camera’s ability to image through zero visibility smoke. This paper analyzes motion and statistical texture features acquired from thermal images to discover the suitable features for accurate classification. Bayesian classifier is implemented to probabilistically classify multiple classes, and a multiobjective genetic algorithm optimization is performed to investigate the appropriate combination of the features that have the lowest errors and the highest performance. The distributions of multiple feature combinations that have 6.70% or less error were analyzed and the best solution for the classification of fire and smoke was identified.

Performing fire endurance tests of railcar floor assemblies in accordance with NFPA 130 is expensive given the minimum size requirements of 3.7 m (12 ft) in length and the full vehicle width. Often it is not financially viable to conduct such tests on several iterations of designs for the purpose of design optimization. Simulations of the fire endurance tests can be performed in place of experiments to provide predictions of floor assembly response of multiple designs at much lower cost. However, capturing the thermo-structural response of the floor assembly requires the ability to model the relevant physical phenomena including softening and weakening of the steel frame, degradation of the fire insulation, and failure of the composite floor. A methodology for performing such simulations was developed under this research addressing each of these phenomena. Temperature dependent thermal and mechanical properties of all modeled materials captures material softening and weakening. Degradation of the insulation was handled through a novel temperature dependent shrinkage approach. Failure models for the sandwich composite floor panels were obtained from literature to predict shear fracture of the core based on a maximum principal shear stress approach and delamination of the core/facesheet based on a maximum strain energy approach. The developed methodology was applied to the simulation of a fire endurance test of an exemplar railcar floor assembly using the commercial finite element solver Abaqus. The assembly was known to hold a passing rating for a 30-minute fire endurance test according to NFPA 130. The floor assembly consisted of a stainless-steel frame, fiberglass insulation, and a ply-metal composite floor. Sequentially coupled thermal and structural models were developed to predict the thermostructural response of the floor assembly for a 30-minute exposure to the ASTM E119 prescriptive fire curve. User-subroutines were utilized to implement the sandwich composite failure models developed for predicted core shear fracture and core/facesheet delamination. The predicted temperature rise on the unexposed surface of the floor assembly after a 30-minute exposure ranged from 50°C to 90°C. The floor assembly was also predicted to maintain structural integrity with the applied crush load, having a center-point vertical deflection of 161 mm after the 30-minute exposure. This resulted in a predicted pass rating for a 30-minute exposure which agrees with the floor assembly’s actual fire rating.

Large scale flammability performance of interior finish used on railcars has been evaluated in previous studies using the NFPA 286 room corner fire test, which has a cross-section similar to a railcar. In some studies, the wall containing the door was removed to account for the shorter length of the room compared to the railcar length. The focus of this study is to assess whether the NFPA 286 standard room-corner test with a door represents conditions that developed inside a railcar during a fire. Fire Dynamics Simulator (FDS) was used to model the fire growth in a NFPA 286 standard room-corner test with a door, NFPA 286 room without the wall containing the door, and railcar geometry with a single door open. All three cases had the same exposure fire in a corner and the same lining material. In predictions of the NFPA 286 room-corner test with a door, gas temperature, heat release rate, and time to flashover agreed well with available NFPA 286 standard test data. The simulation results of fire growth inside a railcar with one side door open produced similar conditions and fire growth compared with the standard NFPA 286 room with a door. For simulations on the NFPA 286 room with the wall containing the door removed, it was found that removal of the wall with the door resulted in non-conservative fire growth conditions with the gas temperature and heat release rate under-estimated compared to the standard NFPA 286 room with a door. These simulations indicate that the standard NFPA 286 room-corner test with a door is representative of conditions that would develop inside of a railcar.

Current standards such as NFPA 130 [1] require railcar floor assemblies to achieve a fire resistance rating according to ASTM E119 [2] by exposing the assemblies to a prescribed 30 minute time-temperature curve using a furnace. Though the ASTM E119 is a standard test procedure, it does not represent a real fire scenario which can have temporal and spatial varying exposure. This work developed a computational framework to evaluate and compare standard fire exposures such as ASTM E119 to real fire exposures to determine the difference in the temperature rise of a railcar floor assembly. The dimensions of the assembly used in this work consisted of the entire width of the railcar ∼3.0 m (10 ft) and a length of 3.7 m (12 ft) as described in NFPA 130. The real fire exposures simulated in this work have been identified in a review [3] of incidents involving fire exposures to railcars in the US and internationally over the past 50 years. The fire exposures consisted of a continuously fed diesel fuel spill, a localized trash fire, and a gasoline spill simulated from a collision of the railcar with an automobile. These realistic fire exposures were applied to a floor assembly model in Fire Dynamics Simulator (FDS) [4] which also included the undercarriage equipment to better capture the fire dynamics. The thermal exposure at the underside of railcar assembly was extracted using the heat transfer coefficient and the adiabatic surface temperature provided by FDS. These spatial-temporal exposures were coupled with a detailed railcar floor assembly finite element (FE) model in ABAQUS [5] to analyze the thermal behavior of the assembly. The thermal model in ABAQUS provided the evolution of temperature in different components of floor assembly consisting of a structural frame, insulation, and a composite floor. The standard scenarios were simulated for two hours instead of the typical 30 minutes to identify the appropriate exposure duration which can better represent a real fire scenario.