Evaluating digital twin data
exchange between a virtual and
physical environment regarding
lighting quantity
Alyaá Tabbah
MASTER THESIS 2021
Master in Product Development with a specialization
SUSTAINABLE BUILDING INFORMATION MANAGEMENT
This thesis work has been carried out at the School of Engineering in Jönköping in the
subject area Product development - Sustainable Building Information Management.
The work is a part of the Master of Science with a major Product development,
specialization in Sustainable Building Information Management.
The authors take full responsibility for opinions, conclusions and findings presented.
Examiner: Myriam Aries
Supervisor: Géza Fischl
Co-Supervisor: Myriam Aries
Scope: 30 credits (second cycle)
Date: 2021/06/22
ADDRESS:School of Engineering, P.O Box 1026, SE-551 11 Jönköping, Sweden VISIT: Gjuterigatan 5, Campus, Building E
PHONE:+46 (0)36 10 10 00 WEB:www.ju.se
Evaluating digital twin data exchange between a virtual and physical
environment regarding lighting quantity
Alyaá Tabbah
Jönköping University, Jönköping 551 11, Sweden taal1696@student.ju.se
Abstract. Building Information Management and Digital Twin technology with help of Smart lights can
opti-mize the built environment impacting our health and well-being, by providing the right amount of light at the right time of day. Lighting simulation is challenging, due to the strict requirements to represent reality. Digital twin technology will provide a more dynamic two-way feed-back between the physical and the virtual environ-ment to optimize the lighting environenviron-ment giving real-time sensor data. The main problem that currently occurs while evaluating a lighting design made in photorealistic computer visualization is using the appropriate form of their model presentation. However, validation of light simulations has been done multiple times but not many studies are based on DT-driven light environment evaluation in which not only the realistic representation but also the exchange of information plays a crucial role. Therefore, the aim is to develop a strategy for demon-strating the data exchange between a physical and real environment, for a scenario in which an optimal inter-action between daylight and electric light derives an optimized realization of a given light demand curve. Based on a quantitative experiment, validation of a Digital Twin was done between a virtual and a physical twin on an existing room using the light simulation tool DIALux evo. Data exchange was optimized for three levels of geometrical complexity. The light environment was optimized for interaction between the Digital and Real Twin. Counter to expectations, the results showed that the coarse model is more accurate representation of the physical counterpart and generates faster data exchange. Defining DT usage purpose reduces time and effort done on the process of creation. Knowing what data to exchange and how often avoid developers any limitations or delaying in the process. Future studies can investigate how optimization of data exchange and light environ-ment can be achieved with programming and parametric generative design.
Keywords: Digital Twin Technology, Physical and Virtual Environment, Data Exchange and Lighting
Envi-ronment
1
Introduction
The built environment impacts our health and well-being on a daily basis through a variety of factors. Light envi-ronment is one factor that mainly influences our visual performance and comfort as well as mood, behavior and interaction with our surroundings (Altomonte et al., 2020). For example, the recommendations of Brown et al., (2020) reveal that rooms with optimized light, both natural and electric, can improve health outcomes such as depression, agitation, and sleep. Building Information Modeling and Management (BIM) gives the AEC industry the opportunity to move forward from traditional working methods to a more digitalized way of working enabling improvement and optimization of the built environment in general and lighting in particular (Siountri, Skondras, Mavroeidakos & Vergados, 2019). Building information modelling will optimize light environment through the application of BIM-uses like Author Design Model, Review Design Model(s), Analyze Lighting Performance (Gerbert et al., 2016; Eastman, Teicholz, Sacks & Lee, 2018). Building information management will use BIM models containing rich information about the building’s assets and light environment to manage and optimize the data exchange and communication between real life and digital BIM models (see Fig. 1) (Siountri, Skondras, Mavroeidakos & Vergados, 2019).
A technology used for design and analysis of lighting is computer simulations even though research showed that these tools are not commonly used by lighting designers and architects (Davoodi, Johansson, Laike & Aries, 2019). A light simulation is a digital representation of the lighting in an existing or planned physical object that does not use any form of automated data exchange between the physical object and the digital object (Kritzinger et al., 2018). Originally, light simulation does not integrate real-time data to cope with regular changes happening in real life, neither can it immediately or directly affect the physical entity (Tao et al., 2018; Lu, 2020).
By implementing Internet of Things (IoT) architectures in the BIM process in different phases of the building life-cycle, 3D virtual models with engaged as-built physical assets could be created as the basis for a new simula-tion approach (Siountri et.al., 2019). This new simulasimula-tion approach is referred to as Digital Twin (DT). In the
2
definition given by Lu, (2020, p.1) a Digital Twin (DT) refers to “a digital replica of physical assets, processes and systems. DTs integrate artificial intelligence, machine learning and data analytics to create living digital sim-ulation models that are able to learn and update from multiple sources, and to represent and predict the current and future conditions of physical counterparts.” Negri et al., (2017) added that sensed data, connected smart devices, mathematical models and real-time data elaboration can optimize lighting design, predictive analytics, self-oper-ating initiatives, and maintenance. The definition is based on preceding research (e.g., Garetti et al., 2012) and is used by many others (e.g., Kritzinger et al., 2018; Tao et al., 2018; Lu, 2020).
According to Kritzinger et al., (2018) a DT can be defined with three levels of integration. The first level is referred to as Digital Model without any automated data exchange. This is equal to the traditional light simulation. The second level implements one-way automated data exchange between the physical and virtual world called Digital Shadow. It only becomes a Digital Twin when there is two-way automated data exchange (see Fig. 1). The real-time data exchange means that sensor data is exported from real life, processed in simulation tool in DT and later optimized giving orders to the lights in real life to change light level. This demands lighting technology (e.g., smart, intelligent, dynamic, or adaptive lighting) in the built environment that can be controlled and communicated with, to adjust lighting according to demand to achieve optimal visual comfort and physical and mental well-being, by providing the right amount of light at the right time of day (Brown et al., 2020; Abd-Alhamid, 2019). More light should be perceived early in the morning and less light exposure in the evening to balance human circadian health (Mead, 2008; Brown et al., 2020). Creating a demand curve that matches the same light exposure pattern will fulfill human lighting demand (social sustainability) (Brown et al., 2020; Casciani & Rossi, 2012). Lighting technology that has compact size, longer lifetime, high efficacy and demand curve that regulates the use of electric light only when needed will save energy (economical and environmental sustainability) (Mead, 2008; Casciani & Rossi, 2012)
Fig. 1. Building Information Modelling and Management framed in purple and green, represent where it is used in DT
con-text. The bidirectional feedback in Digital Twin concept showing third level of integration.
Designing the illumination of a real environment has always been a difficult and often tedious task, requiring much time and effort with the manipulation of physical light sources, shades, reflectors, etc. (Loscos et al., 1999). Light-ing simulation is challengLight-ing, due to the strict requirements to represent reality, but at the same time provide different degrees of complexity for diverse users within the same field (Ochoa, Aries & Hensen, 2012).
The technology of DT enables the simulated virtual environment to be a realistic representation of the experi-ences felt in real physical environments (Abd-Alhamid, 2019). Since DT is a representation of real life it should have high level of detail (LOD) with as built geometric information like size shape, location, quantity and orien-tation and non-geometric information like material, light color temperature, luminous flux (Latiffi, Brahim, Mohd & Fathi, 2015). The realism is especially important in lighting where it has a direct visible effect on the tested environment compared to thermal and sound condition. Parametric BIM objects enable the process of data ex-change to be fully automated in DT. In other word, having one BIM model rich of information that later can be used in different simulation tools depending on the purpose of it, will save time and effort as well as increase realism and accuracy. Every step of transferring data from one model format to another in the process of creating DT may interduce an error factor because of interoperability issues and the way it is performed (Gupta, Cemesova, Hopfe, Rezgui & Sweet, 2014).
3
The huge amount of information contained in BIM models will result into complex virtual environments and the increased number of sensors sending out continuous measurement data result in Big Data sets. The unstructured data requires series of programming techniques to filter it and map it. Big Data Analytics processes relevant data to converted into valuable and understandable information which in principle are tools for building management, optimization and decision-making (Lim, Zheng & Chen, 2020; Ward & Barker, 2013). Visualization of data is often used as a relatively simple way to ensure that more people understand what they see. Instead of having the sensor data measuring the light illuminance in lux in form of a table or grid points, it is more informative to visualize the light distribution in the measured area to show the luminance or brightness received by the eye and possible glare areas. Choosing between reporting illuminance versus luminance data depends on the required out-come. In case of (non-) visual performance of a space, illuminance data are required while for comfort or percep-tion experiences, luminance data are better indicators (Fernandez-Prieto & Hagen, 2017).
However, validation of light simulations has been done already multiple times but not so many studies focus on digital twin-driven lighting simulation evaluation. There is also a challenge in offering DT services in a single environment leading to the fact that some services need a 3D graphic interface and others analyses data only without visualization. The main problem that currently occurs while evaluating a lighting design made in a photo-realistic computer visualization version is using the appropriate form of their model presentation. A fully detailed BIM model has the advantages of realism and as built information but the disadvantages of heaviness and com-plexity that can slow down the data exchange. A simplified BIM model will be easier to manage and serve the simulation purpose. Furthermore, there is a lack in studies that focus on what data should be exchanged in DT context and how often as well as what factors may delay or limit data exchange process.
Therefore, after checking the level of agreement between a virtual (digital) and a physical (reality) twin in terms of lighting quantity (for daylight only), the study investigated how the level of geometrical complexity of the virtual environment impacts time- and error-factors during data exchange with the real environment. Having iden-tified the above research gap, the main aim of this study is to develop a strategy for demonstrating the data ex-change between physical and real environment, for a scenario in which an optimal interaction between daylight and electric light derives an optimized realization of a given light demand curve. The data collected in reality over two days was simulated and compared to each other considering the data exchange management. Different light level alternatives were tested required for an optimal interaction between daylight and electric light in Digital Twin.
2
Methodology
This chapter describes the implementation and workflow of the study. The research strategy is based on quantita-tive case study approach. An experiment was done in SMILE lab, a controlled environment with smart lights, as case study. Daylight illuminance values measured in physical environment were compared to values calculated in virtual environment created using DIALux evo lighting simulation tool. Since the accuracy of the simulation model (the virtual twin) may impact the data exchange, three models with varying geometric complexity (fine, medium, and coarse) were created. This information is used later to check agreement with the amount of light required to meet a given human lighting demand curve. This way, the virtual twin supports a scenario for optimal interaction between daylight and electric lighting to reach an optimized lighting environment. Optimization is reached from a human health perspective, serving right light at the right time and sustainability perspective, using as little energy as possible.
2.1 Experiment
The SMILE lab is a controlled environment in the School of Engineering of Jönköping University, provided with smart light sources and devices. The whole building is located on longitude: 14.16, latitude: 57.78 with a north-east orientation. The SMILE lab has two windows towards the north-north-east allowing direct sunlight to enter the room only in the early hours of the day. The SMILE lab is created to imitate a living environment with both living room and bedroom parts to be able to simulate different scenarios of daily activities. A field experiment was done on SMILE lab since lighting environment is simulated and compared to real-world settings, the definition by Bailey (2008) in experiment design.
4
The photometric indicator to measure light quantity was the illuminance of daylight. The sensed data for illumi-nance values was compared with example demand curve that provides the right amount of light (in lux) at the right time of day to achieve human well-being (see Fig. 2) based on recommendations by Brown et al. (2020). The daylight illuminance and the demand curve illuminance values are constant variables in the experiment that other simulation values were compared to.
Fig. 2. Example of the applied curve for the human lighting demand (horizontal illuminance in lux per hour),as recom-mended by Brown et al. (2020).
2.2 Material
In the physical environment, illuminance was measured using HOBO Pendant MX2202 - Temperature/Light Bluetooth Data Loggers that were distributed systematically in two main grids of total 7 points similar to what presented in Fig. 3. More sensors are concentrated in the living room part since it is the brighter side where most daily routine will occur. The window in the bedroom part is covered with a black-out curtain allowing almost no light to enter from this side. All sensors are placed on horizontal surfaces and measures horizontal illuminance except one placed on the wall measuring vertical illuminance to show the amount of light falling on the vertical area. The sensors are placed at different heights depending on the furniture height. In the virtual environment DIALux evo application was used to provide analysis of the lighting quantity. This software allows following the same procedure described before by placing different measurement points and surfaces.
5
2.3 Procedure
Validation of Digital Twin
The process of creating a DT twin started with using a BIM authoring software application (Revit) based on point cloud file created by photogrammetric laser scanning equipment of the real environment. The fully paramet-ric BIM model with high level of detail was exported in IFC (Industry Foundation Classes) format to the simulation tool (DIALux evo) to perform lighting analysis.
Due to interoperability issues between the two software tools a lot of geometrical information like furniture, light fixtures was lost. Non-geometrical information such as color temperature and luminous flux did not follow to the simulation model as no programming is applied to read from BIM model and map the information automat-ically. This required recreating the model manually in the simulation tool and adding the information needed on the luminaires. The date (March 12th and 17th) and time (between 00:00 and 24:00) in the model were set at the same time measurements were done to show agreement or deviation between the virtual and the physical environ-ment. Location and north alignment were also pre-set as previously mentioned. Next step was adding the meas-urement surfaces. Instead of taking only one point, a surface area was used in order to include the possibility of slightest misplacement of sensor points.
When the simulation model was fully detailed and mapped with real environment, three copies were made with different levels of detail, fine, medium and coarse model in DIALux evo to investigate the impact on data exchange and its characteristics. In results section the different models were referred to with their initials, physical environ-ment (ph), fine model (f), medium model (m) and coarse model (c). The fine model is a very detailed level one Digtial Model that includes as-built objects (Kritzinger et al., 2018). The medium model includes main objects as furniture and openings but complex objects with many surfaces were removed, such as plants, curtains, bed sheet. The coarse model includes main objects but with simplified surfaces as cylinders for circular tables and boxes for rectangular tables and armchairs. In addition, more detail objects were removed, like frames, blackout curtain (see Fig. 4). All levels include the same lighting fixtures and light sources.
Out of seven days of continuous measurements in real life, the sensor data of two days were simulated, March 12th with overcast sky condition and March 17th with clear sky condition and compared to real-life data and demand curve. March 12th was an overcast day but because it was very dark, an alternative clear sky was tested as well. The simulation was limited to 3 surfaces in the beginning, on the window sill, the round coffee table and bed sensors (see Fig. 3), but all sensor points were simulated later to strengthen the accuracy in results (see Appendix 1). The ratio between the simulated data in all three models and real-life data was calculated. The results presented in next chapter includes data sets for one sensor point on the round table in the center of the living room, for full calculations of all sensor points see Appendix 1.
Fig. 4. Three geometrical complexity levels of the simulation model respectively, fine, medium and coarse.
Optimization of data exchange
The data sets of all sensor points on the two measurement days were mapped in one Excel file and the average time for running simulations in hourly-steps of the three models was calculated manually to see the relation with the detail-factor. Additionally, the influence of building height from surrounding buildings was investigated. Two heights were tested: h= 15.4 m and h= 10.5 m to check if the shadow of the facing building effects the amount of light entering the SMILE lab. Finally, the option in DIALux evo that calculates only calculation surfaces was tested to check how much time it reduces compared to entire project calculation with all surfaces.
6
Optimization of light environment
The final step was demonstrating how Digital Twin (DT) and Real Twin (RT) work together to reach the optimal solution for realizing a lighting demand curve (see Figure 2) at one chosen point in the space (sensor point on the round table in the center of the living room) (see Fig. 3). Based on the agreement of simulation data and demand curve data meeting the human needs, the difference between demand curve illuminace values and simulated values was calculated and by that the required amount of electric lighting was determined. The same simulation was run predicting 5 minutes in future. The DT calculated the optimal luminous flux needed to reach demand values but at the same time not exceeding it and waste energy. This was done by turning on electric light source in different capacity levels and testing it.
3
Results
This chapter presents the main results by first describing the validation between the Real and Digital Twin and subsequently related to the optimization of consequences for the data exchange and lastly optimization of light environment.
3.1 Validation of Digital Twin
The results generated for daylight illuminance in the simulation tool were compared to real-time sensor data. In both charts of Fig. 5 only one measurement sensor point was presented to visualize the chart in a clear way which is the sensor point on the round table in the center of the living room in SMILE lab (see Fig. 3). It was chosen because it does give the most agreed values with sensor data and demand curve and it is the area where most likely activities will occur, like desk work or daily routine. Fig. 5.a represents comparison of real-life data for daylight illuminance values from the physical environment with all three models. This showed that the simulated data does not agree with the sensor data except in the morning hours between 6-9 a.m. because of the location of building. Fig. 5.a shows that both the real-time sensor data and the simulated data do not reach the demand curve level as well. Fig. 5.b represents a comparison of the two measurement days March 12th with overcast sky and March 17th with clear sky. The chart shows that there is a significant increasing of daylight illuminance values depending on the sky condition but the same pattern of light distribution during the day. The clear sky allows more light to reach the room. Both charts in Fig. 5 shows a tendency of more morning light and less afternoon light.
(a) (b)
Fig. 5. (a) comparison of real-life data in physical environment (ph) and simulation data from the fine model (f), medium
model (m) and coarse model (c). (b)comparison of the two measurement days March 12th with overcast sky and March 17th
with clear sky.
Focusing on the center of the living room, the ratio between the simulated data and the sensor data was calcu-lated. The average agreement between the environments is then generated through each simulation model. Table 1. concludes that the fine model has an agreement of 30%, medium model has 39% and the coarse model has 40%.
7
The average ratio demonstrates that the coarse model has the best agreement therefore it was used to add the electric light. For full calculations of all sensor points see Appendix 1.
Table 1. Raw illuminance data for physical environment and all three models calculated into the ratio between the
simu-lated data and the sensor data for sensor point on the round table in the living room center.
3.2 Optimization of data exchange
After applying the three levels geometrical complexity in DIALux evo the average simulation time was calcu-lated manually to check how it was affected. Table 2. summarizes the results for this step. It includes the two measurement days March 12th and 17th, the three levels of geometrical complexity fine, medium and coarse model as well as how the calculation was done, for the entire project and surfaces or for calculation surfaces only. The average time of calculation for March 17th with clear sky condition and higher illuminance values takes slightly longer time compared to March 12th with overcast sky condition. The coarse model has a significantly faster simulation time when the calculation is done for the entire project and surfaces, followed by the medium model and last the fine model. This demonstrates that the level of geometrical complexity will affect the time factor in a positive relationship graph. As expected, the higher the level of complexity the longer time it takes to simulate. When using the option of calculating only calculation surfaces where the real sensors are placed, the average simulation time drops very much in all three models. This can be a choice when illuminance results will only be presented as values and not for visualization purpose or rendering. Both options “entire project calculation” and “calculation surfaces only” give same illuminance values at the same measurement point and date/time. This means that “calculation surfaces only” option has high accuracy and faster results.
Table 2. The influence of the granularity of the model on the average simulation time in fine, medium and coarse model for
the entire project option and the calculation surfaces only option in the two measurement days.
Date Time, GMT +0100 ph- illuminance f-model illuminance m-model illuminance c-model illuminance f/ph- ratio m/ph- ratio c/ph- ratio 2021-03-12 06:00 0 0 0 0 0,0% 0,0% 0,0% 2021-03-12 07:00 22 33,5 39,8 40,4 153,1% 181,9% 184,6% 2021-03-12 08:00 112 243 352 363 216,2% 313,2% 323,0% 2021-03-12 09:00 213 277 405 411 130,2% 190,4% 193,2% 2021-03-12 10:00 322 113 132 132 35,1% 41,0% 41,0% 2021-03-12 11:00 242 95,2 109 109 39,3% 45,0% 45,0% 2021-03-12 12:00 194 79,2 89,2 90 40,9% 46,1% 46,5% 2021-03-12 13:00 333 66,6 74,4 75,1 20,0% 22,3% 22,5% 2021-03-12 14:00 218 58,8 64,2 64,1 26,9% 29,4% 29,4% 2021-03-12 15:00 265 49,6 54,5 54,8 18,7% 20,6% 20,7% 2021-03-12 16:00 141 39,3 42,8 42,5 28,0% 30,4% 30,2% 2021-03-12 17:00 40 23 25,5 25,4 57,4% 63,6% 63,4% 2021-03-12 18:00 2 0 0 0 0,0% 0,0% 0,0% 2021-03-12 19:00 0 0 0 0 0,0% 0,0% 0,0% 2021-03-12 20:00 0 0 0 0 0,0% 0,0% 0,0% 30,6% 39,4% 40,0% average accuracy
Data for sensor point on the round table in the living room center
Entire project calculation Calculation surfaces only Entire project calculation Calculation surfaces only Entire project calculation Calculation surfaces only March 12th average time 04:54 00:29 01:59 00:23 01:00 00:20 March 17th average time 05:13 00:30 02:04 00:26 01:00 00:20 Calculation time Calculation time Calculation time coarse model medium model fine model
8
As the exact height of the surrounding building of SMILE lab room was not known, a simple model was created in DIALux evo with two building heights for the surrounding buildings over a set of time points. Table 3. showed that the decreased building height will allow increased light received in SMILE lab. In the represented set of points, illuminance increased by 12-33%.
Table 3. A model with two total surrounding building heights over a set of points shows an inverse relationship between
height and received light amount
3.3 Optimization of light environment
This part demonstrates how the Digital Twin and Real Twin work together to reach the optimal solution for realizing the lighting demand curve at one chosen point in the space (the sensor on the round table in the living room). Table 4. shows the difference in illuminance values between simulation data and demand curve on March 12th. In all models the difference is negative except the values marked in red. This means at this time point there is no need for electric light. By knowing how much light is needed, the electric light can be determined. In Table 4. both daylight and electric light were calculated to reach demand curve level. The first set was calculated by turning on all light sources that actually exist in real-life in the living room to maximum intensity. Even though, this amount of light was not enough to achieve demand curve so two brighter task lights were simulated and turned on in different (dimming) levels. This data set does fulfill the demand as shown in Table 4.
Table 4. Data sets on March 12th representing difference in between simulated values and demand curve determining how
much electric light is needed for both warm light and task light (in the living room where the sensor on the round table is)
Next step was 5 minutes future prediction simulation with the use of task light in the coarse model (see Table 5). This data set holds constant level of light except the red marked value. When the DIALux evo simulated +5 min at 16:00 with same light level as before 35% (see Table 4), the illuminance value (298 lux) got lower than preset demand value (300 lux). This required and increased light level to 40% reaching (355 lux). In Table 5. the light level increased/decreased with 5% steps when needed but since the calculation can be generated very fast in the coarse model, light level can go down to 1% steps even before the next +5 minutes occur. The DT can calculate
15,4 m 10,5 m 09:00:00 201 225 12% 10:00:00 85 110 29% 11:00:00 75 98 31% 12:00:00 66 88 33% Illuminance, lux height /time Increasing Date Time, GMT +0100 Demand illuminance f-model illuminance m-model illuminance c-model illuminance f-demand difference m-demand difference c-demand difference Warm light 100% on, lux Task light, lux Turned On 2021-03-12 06:00 50 0 0 0 -50,0 -50,0 -50,0 90 73,0 10% 2021-03-12 07:00 150 33,5 39,8 40,4 -116,5 -110,2 -109,6 131 151,0 15% 2021-03-12 08:00 300 243 352 363 -57,0 52,0 63,0 363 363 0% 2021-03-12 09:00 300 277 405 411 -23,0 105,0 111,0 411 411 0% 2021-03-12 10:00 450 113 132 132 -337,0 -318,0 -318,0 222 462,0 45% 2021-03-12 11:00 450 95,2 109 109 -354,8 -341,0 -341,0 200 477,0 50% 2021-03-12 12:00 450 79,2 89,2 90 -370,8 -360,8 -360,0 180 457,0 50% 2021-03-12 13:00 450 66,6 74,4 75,1 -383,4 -375,6 -374,9 166 479,0 55% 2021-03-12 14:00 300 58,8 64,2 64,1 -241,2 -235,8 -235,9 155 321,0 35% 2021-03-12 15:00 300 49,6 54,5 54,8 -250,4 -245,5 -245,2 145 312,0 35% 2021-03-12 16:00 300 39,3 42,8 42,5 -260,7 -257,2 -257,5 133 300,0 35% 2021-03-12 17:00 300 23 25,5 25,4 -277,0 -274,5 -274,6 116 319,0 40% 2021-03-12 18:00 150 0 0 0 -150,0 -150,0 -150,0 90 184,0 25% 2021-03-12 19:00 150 0 0 0 -150,0 -150,0 -150,0 90 184,0 25% 2021-03-12 20:00 50 0 0 0 -50,0 -50,0 -50,0 90 73,0 10%
9
first what happens if real light sources are turned on 100%. If the electric light together with the actual daylight situation and the situation 5 minutes later exceed the demand curve, then the electric lighting can be dimmed and the DT calculates the results again for electric light at 50%, and so on until it reaches the optimal light condition according to demand curve.
Table 5. Data sets representing 5 min future prediction simulation using task light in the coarse model for sensor point on the
round table in the living room center.
4
Discussion
The study was conducted in three main part. First, evaluating if the DT is good representation of the Real Twin. Second, optimizing the data exchange through the presentation of different level of geometrical complexity, fine, medium and coarse model. The third part and the core of the study is generating and analyzing a scenario of optimal interaction between daylight and electric light to reach optimized realization of a given light demand curve and meet the human lighting needs in different times of the day. This will contribute in a systematic way of creat-ing, evaluating and optimizing DT based on the purpose it serves. In this study, Digital Twin can be achieved in the sense of manipulating the virtual environment in the simulation tool (Digital Model) manually instead of au-tomatic programmed data exchange according to the Level of Integration (Kritzinger et al., 2018).
4.1 Validation of Digital Twin
Regarding the evaluation, the simulation data does not agree with real life data because the simulation tool cannot represent the complex light condition in real life. The simulation model assumes that the sky is overcast through the whole day while the curve of real-life shows variations in the illuminance. This was a limitation in the simulation tool where the sky horizontal illuminance cannot be controlled or mapped with the regularly changing real-time data. Another reason for the disagreement is the unknown surrounding conditions of the building height. This shows the importance of having as-built BIM models (Latiffi, Brahim, Mohd & Fathi, 2015). If this infor-mation was available, it would limit the chance of introducing an error factor in the process of creating DT.
In facility management phase, analyses of building performance are usually done based on BIM models re-ceived from previous phases. To avoid doubling the work of recreating the model in the simulation tool the data can be exchanged automatically between the BIM model and simulation model. In this case study, the BIM tool and the simulation tool had interoperability issues that cannot be controlled resulting in doubling the amount of work, time and effort to recreate a simulation model rich of information. Furniture objects and lighting fixtures that were placed in Revit did not appear in DIALux evo so they had to be replaced in the simulation tool. Even when these objects were imported directly in DIALux evo from other software tools like SketchUp, the dimensions
Date Time, GMT +0100
Demand illuminance
5+ min with task light, lux
Turned On 2021-03-12 06:00 50 73,0 10% 2021-03-12 07:00 150 162,0 15% 2021-03-12 08:00 300 363,0 0% 2021-03-12 09:00 300 411,0 0% 2021-03-12 10:00 450 461,0 45% 2021-03-12 11:00 450 474,0 50% 2021-03-12 12:00 450 456,0 50% 2021-03-12 13:00 450 477,0 55% 2021-03-12 14:00 300 320,0 35% 2021-03-12 15:00 300 311,0 35% 2021-03-12 16:00 300 355,0 40% 2021-03-12 17:00 300 318,0 40% 2021-03-12 18:00 150 184,0 25% 2021-03-12 19:00 150 184,0 25% 2021-03-12 20:00 50 73,0 10%
Data for sensor point on the
10
were distorted (very big) and needed manual manipulation. The simulation tool did not support other file formats and had limited library of objects. This is one of the challenges faced when implementing a new technology in traditional working methods (Siountri etal., 2019). In most cases BIM systems at companies are equipped with traditional tools for light simulation (e.g. DIALux evo, Velux, Revit Insight 360). These tools need to be updated and completed with smart devices (e.g. sensors, weather station and Smart light sources) and programming (to connect the physical world with virtual world) to cope with DT technology (Macchi et al., 2018).
4.2 Optimization of data exchange
Regarding the optimization of data exchange, the lower the level of geometrical complexity is the faster the simulation is in terms of the time factor. Unexpectedly, the lower level of geometrical complexity showed better agreement with real world and demand curve. The coarse model was mostly developed directly in DIALux evo. This can be due to errors introduced when transferring data to different file formats. One of the windows in the transferred model was not recognized by the simulation tool when imported as BIM model. Replacing the imported window model manually with a DIALux window model worked with the light simulation. The average time of calculation for March 17th with clear sky condition and higher illuminance values takes slightly longer time com-pared to March 12th with overcast sky condition. This probably has to do with the fact that a clear sky is a combi-nation of direct and diffuse light calculations while a model using an overcast sky has only diffuse light calcula-tions. The BIM-uses applied in this project were Author Design Model and Review Design Model in Revit and Analyze Lighting Performance in DIALux evo. These Building Information Modelling-uses will optimize light environment in the Digital Twin (see Fig. 1) (Gerbert et al., 2016; Eastman, Teicholz, Sacks & Lee, 2018). Build-ing Information Management will use the Revit model containBuild-ing rich information about the buildBuild-ing’s assets and light environment to manage and optimize the data exchange in DIALux evo and communication between RT and DT (see Fig. 1) (Siountri, Skondras, Mavroeidakos & Vergados, 2019).
4.3 Optimization of light environment
Regarding the optimal interaction between the Real and Digital Twin for creating an optimal combination of daylight and electric lighting, Digital Twin plays a crucial rule to find the optimal point for human well-being and energy saving. The task being performed requires different lighting conditions. Reading as activity requires higher illuminance than laying down for example. Therefore, it is important to use appropriate demand curve when pre-senting different scenarios of daily activities. The place may also affect the purpose of the demand curve because you do curtain activities in curtain places. By knowing the difference in illuminance values between simulation data and demand curve, the electric light can be determined and therefore energy consumption minimized. At times where daylight is sufficient in achieving human light demand, electric lighting can be turned off to save more energy. For example, on March 12th between 8:00-9:00 daylight was sufficient. Turning off two task lights for two hours (e.g. 60 Watts*2 lights*2 hours) will save 240 Wh per day. When there is need for electric light the DT will turn on electric light sources in different capacity levels to reach the optimal luminous flux for demand values but at the same time not exceed it and waste energy (Casciani & Rossi, 2012). On March 12th at 10:00, electric light sources needed to be turned on 45% in addition to daylight. An hour later the DT simulated that the same light level was not enough so electric light capacity was increased to 50%.
When electric light is simulated and turned on in different degrees, the demand curve is fulfilled. This means that more light was provided in the morning and less in the evening by combining electric light with daylight (Brown et al., 2020). The person sitting in the room will feel awake and refreshed to perform tasks (like writing) in an environment provided with sufficient light for visual comfort (Mead, 2008). If demand curve is not used and the light sources are turned 100% on constantly during the day, the person might experience too bright surrounding reflecting a lot of light and possibly glare that will make the eyes tired. Getting exposed for light in the evening for long time will result in late bed-time (Mead, 2008). This will also consume a lot of energy per day (Casciani & Rossi, 2012). Good light distribution is also important to avoid contrasting surrounding with very bright and dark areas that exhausts the eyes (Mead, 2008). People sitting in this environment might turn on extra light sources to get a better distribution which will mean extra energy consumption. If electric light sources are turned on lower than demand curve, the person will experience tiredness and sleepiness during the day making him unable to sleep
11
at night (Mead, 2008; Brown et al., 2020). The eyes will be tired of focusing in the dark when performing detail tasks like writing. The energy consumption might be lowered during the day but the person will most probably turn on more lights during the night because of alertness.
The quick simulation time and the continues measurements of real-time data allows the digital twin to test and control different alternatives in the virtual environment before sending information to real-life predicting what may happen in the near future. For example, when DT simulated 5 minutes in future at 16:00, electric light was not enough at 35%. The DT then simulated 36%, 37%, 38%, 39% and 40% to reach demand curve with the optimal light level. When DT predict a need of increasing light level, it will send information to RT to turn on electric light gradually brighter within the 5 minutes to reach better visual comfort. Instead of testing each light level in DT and sending information to RT, testing all different light level alternatives in virtual world and finally choosing the best alternative to send it once to Real Twin will save a lot of energy. This process can only be done with future prediction.
4.4 Limitations
The overall method described was very simplified. The study was limited to studying only the light environ-ment since all disciplines cannot be examined in one paper. Lighting was of more interest as it has direct and immediate effect that can be seen compared to thermal and sound conditions. The focus was only on SMILE lab as controlled environment because it was provided with Smart lighting that help optimize energy and human well-being. This is where the demand curve is required to give people the right light at the right time of the day. The Lighting environment, SMILE lab room and the demand curve are just examples in the process of optimizing digital twin. The main point is that the same process could be done on any type of building or discipline by fol-lowing the same procedure (strategy). The measurements were narrowed down by the limitation to one photometric indictor and a few sets of sensor points. This enabled in the study to test different variations of the Digital Model but all in a very basic and shallow scope. The different surrounding building heights, task light and warm light, only calculation surfaces option, an overcast and clear sky model, fine medium and coarse level of detail models and different simulation days, all are variations worth studied individually to formulate a deeper understanding of its effects on the Data-exchange in DT context. The experiment was limited to one orientation, one room, no repeated measurement or no quality data regarding unobstructed daylight conditions in order to apply the correct sky model.
5
Conclusions
Researchers and practitioners are becoming more interested of the DT in industry in recent years. This paper contributes to the body of knowledge by presenting a digital twin-driven lighting simulation and optimized char-acteristics of data-exchange between physical and virtual environment. To help achieve the optimized realization of real environment we need to limit the error factor in the sensing equipment and the measurement way that can create deviation from the real life. Therefore, it is recommended to define the purpose of why the Digital Twin is developed in the beginning to reduce the time and effort done on the process of creation. Having this in mind then developer can know what data should be exchanged, how it should be exchanged and how often to reach an optimal environment avoiding any limitations or delaying the process. This will also set requirements on the simulation tools used in future studies of DT-driven simulations, as being flexible, controllable, programable and interopera-ble with different data input.
Future studies can investigate to what extend error factor should be minimized to give an accurate representation of the physical environment. Implementing programming, parametric design and generative design is an interest-ing method for optimizinterest-ing any process in the DT. Data exchange, light distribution and the visual comfort can be optimized by this technology. Many papers present DT connectivity and architecture techniques out of program-ming and technical perspective. On the other hand, BIM-users, architectures and engineers that will implement DT systems in their projects to benefit from maintenance and operations monitoring and optimization, do not always understand what it consists of, how it is connected and how the communication between all parts is per-formed. Therefore, future studies should focus on implementing “BIM and IoT devices integration methods” from the BIM-user or even the end-user perspective.
12
References
Abd-Alhamid, K. (2019). Developing an Innovative Method for Visual Perception Evaluation in a Physical-Based Virtual Environment. Building and Environment, 162, 106278–. https://doi.org/10.1016/j.buildenv.2019.106278
Altomonte, S., Allen, J., Bluyssen, P. M., Brager, G., Heschong, L., Loder, A., Schiavon, S., Veitch, J. A., Wang, L., & Wargocki, P. (2020). Ten questions concerning well-being in the built environment. Building and Environment, 180, 106949– . https://doi.org/10.1016/j.buildenv.2020.106949
Bailey, R.A. (2008). Design of Comparative Experiments. Cambridge: Cambridge University Press. ISBN 9780521683579
Brown, T., Brainard, G., Cajochen, C., Czeisler, C., Hanifin, J., Lockley, S., Lucas, R., Munch, M., O'Hagan, J., Peirson, S., Price, L., Roenneberg, T., Schlangen, L., Skene, D., Spitschan, M., Vetter, C., Zee, P., Wright Jr., K. (2020). Recommen-dations for Healthy Daytime, Evening, and Night-Time Indoor Light Exposure. Preprints 2020, 2020120037. doi:10.20944/preprints202012.0037.v1
Casciani, D. & Rossi, M. (2012). ELSE, Experience of Lighting Sustainability in the Environment. Cumulus Northern World
Mandate conference.
https://www.researchgate.net/publication/308948048_ELSE_Experience_of_Lighting_Sustainability_in_the_Environment
Cimino, C., Negri, E., & Fumagalli, L. (2019). Review of digital twin applications in manufacturing. Computers in
Indus-try, 113, 103130–. https://doi.org/10.1016/j.compind.2019.103130
Davoodi, A., Johansson, P., Laike, T., Aries, M. (2019). Current use of lighting simulation tools in Sweden. Linköping: Linköping University Electronic Press, 60th International Conference of Scandinavian Simulation Society, SIMS 2019, Väs-terås, Sweden, August 13-16, 2019
Eastman, C. M., Teicholz, P. M., Sacks, R., & Lee, G. (2018). BIM handbook: a guide to building information modeling for
owners, managers, designers, engineers and contractors (Third edition). Wiley
Fernandez-Prieto, D., & Hagen, H. (2017). Visualization and Analysis of Lighting Design Alternatives in Simulation Soft-ware. Applied Mechanics and Materials, 869, 212–225. https://doi.org/10.4028/www.scientific.net/AMM.869.212
Garetti, M., Rosa, P., Terzi, S., (2012). Computers in industry life cycle simulation forthe design of product – service sys-tems. Comput. Ind. 63, 361–369, http://dx.doi.org/10.1016/j.compind.2012.02.007
Gerbert, P., et al. (2016). Shaping the Future of Construction – a Breakthrough in Mindset and Technology, World Eco-nomic Forum in collaboration with The Boston Consulting Group, Cologny/Geneva Switzerland.
Gupta, A., Cemesova, A., Hopfe, C. J., Rezgui, Y., & Sweet, T. (2014). A conceptual framework to support solar PV simulation using an open-BIM data exchange standard. Automation in Construction, 37, 166–181. https://doi.org/10.1016/j.autcon.2013.10.005
Kritzinger, W., Karner, M., Traar, G., Henjes, J., & Sihn, W. (2018). Digital Twin in manufacturing: A categorical literature review and classification. IFAC PapersOnLine, 51(11), 1016–1022. https://doi.org/10.1016/j.ifacol.2018.08.474
Latiffi, A., Brahim, J., Mohd, S., & Fathi, M. (2015). Building Information Modeling (BIM): Exploring Level of Develop-ment (LOD) in Construction Projects. Applied Mechanics and Materials, 773-774, 933–937. https://doi.org/10.4028/www.sci-entific.net/AMM.773-774.933
Lim, K., Zheng, P., & Chen, C. (2020). A state-of-the-art survey of Digital Twin: techniques, engineering product lifecycle management and business innovation perspectives. Journal of Intelligent Manufacturing, 31(6), 1313–1337. https://doi.org/10.1007/s10845-019-01512-w
13 Loscos, C., Frasson, M. C., Drettakis, G., Walter, B., Granier, X., Poulin, P. (1999). Interactive Virtual Relighting and Remodeling of Real Scenes In D. Lischinski and G.Ward-Larson, (eds), Rendering Techniques ’99 (10th EG workshop on Rendering). Springer Wien.
Lu, P. (2020). Developing a Digital Twin at Building and City Levels: Case Study of West Cambridge Campus. https://doi.org/10.17863/CAM.45198
Macchi, M., Roda, I., Negri, E., Fumagalli, L., (2018). Exploring the role of digital twin for asset lifecycle management. IFAC-Papers On Line. 51, 790–795, http://dx.doi.org/10.1016/j.ifacol.2018.08.415
Mead, M. N. (2008). Benefits of Sunlight: A Bright Spot for Human Health. Environmental Health Perspectives, 116(4), A160–A167. https://doi.org/10.1289/ehp.116-a160
Negri, E., Fumagalli, L., Macchi, M., (2017). A review of the roles of digital twin in CPS-based production systems. Pro-cedia Manuf. 11, 939–948, http://dx.doi.org/10.1016/j.promfg.2017.07.198
Ochoa, C., Aries, M., & Hensen, J. (2012). State of the art in lighting simulation for building science: a literature re-view. Journal of Building Performance Simulation, 5(4), 209–233. https://doi.org/10.1080/19401493.2011.558211
Santos, E. & Derani, L. (2003). An Immersive Virtual Reality System for Interior and Lighting Design. Proceedings of the 8th International Conference on Computer Aided Architectural Design Research in Asia. ISBN 974-9584-13-9. pp. 593–596. http://papers.cumincad.org/cgi-bin/works/paper/caadria2003_c4-1
Siountri, K., Skondras, E., Mavroeidakos, T., & Vergados, D. (2019). The Convergence of Blockchain, Internet of Things (IoT) and Building Information Modeling (BIM): The smart museum case.
Tao, F., Cheng, J., Qi, Q., Zhang, M., Zhang, H., Sui, F., (2018). Digital twin-driven product design, manufacturing and service with big data. Int. J. Adv. Manuf. Technol. 94,3563–3576, http://dx.doi.org/10.1007/s00170-017-0233-1
Veitch, J.A., Davis, R.G. (2019). Lighting Research Today: The More Things Change, the More They Stay the Same. LEUKOS J. Illum. Eng. Soc. N. Am., 15, 77–83. https://doi.org/10.1080/02604027.2019.1596680
Appendix 1: All calculation data for all sensor points in the two measurment days March 12th nad 17th
Intensity, lux fine colored clear medium colored coarse colored
Date Time, GMT +0100 ph-bed ph-shelf ph-kitchen ph-round tableph-rect tableph-frame ph-windowDemand Calc time Surface onlyf-bed f-shelf f-kitchen f-round tablef-rect table f-frame f-window window oldCalc time Surface onlym-bed m-shelf m-kitchen m-round tablem-rect tablem-frame m-window window oldCalc time Surface onlyc-bed c-shelf c-kitchen c-round tablec-rect tablec-frame c-window window old
2021-03-12 0:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-12 1:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-12 2:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-12 3:00:00 0 0 0 0 0 0 0 1 03:59,8 0 0 0 0 0 0 0 0 01:48,9 0 0 0 0 0 0 0 0 00:48,1 0 0 0 0 0 0 0 0 2021-03-12 4:00:00 0 0 0 0 0 0 0 1 04:06,7 0 0 0 0 0 0 0 0 01:47,0 0 0 0 0 0 0 0 0 00:55,7 0 0 0 0 0 0 0 0 2021-03-12 5:00:00 0 0 0 0 0 0 0 1 04:02,1 0 0 0 0 0 0 0 0 01:47,5 0 0 0 0 0 0 0 0 00:53,3 0 0 0 0 0 0 0 0 2021-03-12 6:00:00 0 0 0 0 0 0 1 50 04:06,8 00:26,2 0 0 0 0 0 0 0 0 01:47,6 00:20,7 0 0 0 0 0 0 0 0 00:53,8 00:19,3 0 0 0 0 0 0 0 0 2021-03-12 7:00:00 2 16 8 22 11 16 311 150 05:19,0 00:30,2 4,28 51,9 8,15 33,5 13,4 22,8 222 147 02:14,2 00:27,1 7,55 53,6 11,5 39,8 18,3 26,1 231 147 01:05,6 00:22,0 6,88 53,6 11,4 40,4 19,2 26,5 229 229 2021-03-12 8:00:00 15 75 42 112 60 96 1 716 300 05:15,5 00:31,8 27,4 140 88,3 243 145 287 1331 1269 02:06,3 00:26,2 69,7 165 131 352 208 330 1472 1266 01:07,8 00:25,6 61,1 167 142 363 231 336 1447 1447 2021-03-12 9:00:00 26 130 83 213 128 190 4 765 300 05:15,5 00:28,9 34,5 155 118 277 169 265 4581 643 02:07,9 00:24,6 64,9 179 158 405 245 361 4707 643 01:11,9 00:20,6 56,9 197 165 411 259 365 4670 4670 2021-03-12 10:00:00 37 180 129 322 227 338 2 900 450 05:11,9 00:27,4 9,97 73,7 24 113 43 58,8 1095 634 02:03,1 00:22,2 19,3 79,6 34,4 132 57,4 67,4 1124 634 01:02,6 00:18,6 16,7 78,7 34,6 132 59,3 68,2 1111 1111 2021-03-12 11:00:00 22 121 81 242 118 152 2 646 450 05:09,4 00:29,8 7,51 54,2 18 95,2 34,6 44,8 939 573 02:01,1 00:22,7 14,2 57,6 24,8 109 47,4 51,9 960 573 01:00,8 00:18,7 12,7 57,5 25,8 109 48,5 53,9 950 950 2021-03-12 12:00:00 17 89 60 194 85 105 2 159 450 05:06,3 00:28,5 5,81 41,1 14,3 79,2 28,3 36,3 775 504 01:56,7 00:22,9 10,4 43,1 19,3 89,2 37,6 42,9 791 504 00:59,0 00:18,8 9,74 43,5 19,2 90 39,3 43 782 782 2021-03-12 13:00:00 23 173 82 333 142 149 2 740 450 05:25,0 00:28,2 4,6 33,3 11,5 66,6 24,9 29,8 645 442 02:03,6 00:23,1 8,34 35,9 15,5 74,4 32,9 32,7 657 442 01:01,3 00:18,3 7,31 35,9 14,4 75,1 34,1 33 651 651 2021-03-12 14:00:00 16 99 56 218 87 100 2 075 300 05:17,1 00:28,0 3,71 29,6 9,33 58,8 21,6 22,3 540 384 02:02,2 00:21,9 6,94 31,5 12,8 64,2 28,7 25,8 551 384 01:00,9 00:17,9 6,23 31,2 12,8 64,1 28,3 25,9 545 545 2021-03-12 15:00:00 14 114 50 265 98 83 1 456 300 05:18,6 00:28,4 3,38 25,6 8,08 49,6 18,5 20,5 443 323 02:00,4 00:23,1 6,07 27,2 10,6 54,5 24 23,7 451 323 01:00,2 00:18,2 5,25 27,1 11 54,8 24,7 22,7 447 447 2021-03-12 16:00:00 9 59 33 141 52 53 1 021 300 05:20,9 00:28,8 2,7 20,8 6,29 39,3 14,2 14,2 335 247 02:03,0 00:22,7 4,63 22,1 8,24 42,8 18,8 16,4 342 247 00:59,1 00:18,3 4,23 22,1 8,17 42,5 19,2 17,1 338 338 2021-03-12 17:00:00 3 18 9 40 15 16 286 300 05:20,3 00:29,6 1,61 12,9 3,59 23 8,39 8,61 194 143 02:01,5 00:22,9 2,76 13,4 4,68 25,5 11,3 10,1 198 143 01:01,1 00:18,5 2,55 13,4 4,96 25,4 11,5 9,68 196 196 2021-03-12 18:00:00 0 1 1 2 1 1 29 150 04:12,2 0 0 0 0 0 0 0 0 01:48,8 0 0 0 0 0 0 0 0 00:51,5 0 0 0 0 0 0 0 0 2021-03-12 19:00:00 0 0 0 0 0 0 0 150 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-12 20:00:00 0 0 0 0 0 0 0 50 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-12 21:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-12 22:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-12 23:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-13 0:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 average 04:54,2 00:28,8 01:58,7 00:23,3 00:59,6 00:19,6 2021-03-17 0:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 1:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 2:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 3:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 4:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 5:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 6:00:00 0 0 0 1 0 1 12 50 04:12,2 00:29,2 0 0 0 0 0 0 0 0 01:47,7 00:25,0 0 0 0 0 0 0 0 0 00:51,9 00:19,3 0 0 0 0 0 0 0 0 2021-03-17 7:00:00 3 15 9 27 12 17 397 150 05:22,5 00:30,2 5,56 59,1 11,55 46,8 20,7 46,2 278 209 02:06,1 00:27,1 9,91 60,8 14,9 55 25,6 49,5 287 210 01:01,7 00:22,0 9 60,8 14,8 56,2 26,5 49,9 285 322 2021-03-17 8:00:00 8 38 28 86 40 52 1 252 300 05:22,3 00:31,8 33,4 147,2 91,7 319 152,3 310,4 1387 1866 02:11,8 00:27,2 78,1 172,2 134,4 463 215,3 353,4 1528 1862 01:03,9 00:25,6 71 174,2 145,4 482 238,3 359,4 1503 2084 2021-03-17 9:00:00 71 245 207 452 332 523 14 746 300 05:20,6 00:30,2 39 162,2 121,4 312 176,3 288,4 4637 751 02:10,1 00:26,6 72,1 186,2 161,4 459 252,3 384,4 4763 751 01:03,1 00:20,6 62,8 204,2 168,4 465 266,3 388,4 4726 5812 2021-03-17 10:00:00 43 155 127 262 231 384 1 727 450 05:13,1 00:29,5 10,4 80,9 27,4 122 50,3 82,2 1151 685 02:00,8 00:25,8 20,3 86,8 37,8 143 64,7 90,8 1180 685 01:00,2 00:18,6 17,9 85,9 38 142 66,6 91,6 1167 1199 2021-03-17 11:00:00 50 239 251 540 351 474 2 237 450 05:17,7 00:29,8 7,86 61,4 21,4 102 41,9 68,2 995 608 02:02,4 00:25,7 14,5 64,8 28,2 116 54,7 75,3 1016 608 01:00,6 00:18,7 13,1 64,7 29,2 116 55,8 77,3 1006 1005 2021-03-17 12:00:00 31 266 117 647 361 246 2 534 450 05:17,4 00:29,6 5,9 48,3 17,7 82,3 35,6 59,7 831 529 02:03,8 00:25,7 10,8 50,3 22,7 93,3 44,9 66,3 847 529 00:59,9 00:18,8 9,75 50,7 22,6 93,6 46,6 66,4 838 821 2021-03-17 13:00:00 31 288 94 1 416 307 186 2 072 450 05:15,9 00:29,7 4,72 40,5 14,9 69,6 32,2 53,2 701 462 02:00,5 00:25,5 9,05 43,1 18,9 77,1 40,2 56,1 713 462 00:59,1 00:18,9 7,74 43,1 17,8 76,9 41,4 56,4 707 679 2021-03-17 14:00:00 6 48 20 138 43 35 567 300 05:14,1 00:29,9 4,09 36,8 12,73 60,4 28,9 45,7 596 402 02:12,8 00:25,8 7,51 38,7 16,2 66,9 36 49,2 607 402 00:59,5 00:18,6 6,35 38,4 16,2 67 35,6 49,3 601 570 2021-03-17 15:00:00 11 121 42 290 112 66 1 148 300 05:18,1 00:29,5 3,6 32,8 11,48 51,8 25,8 43,9 499 341 02:01,9 00:25,6 6,23 34,4 14 57,3 31,3 47,1 507 341 00:59,6 00:18,2 5,55 34,3 14,4 56,8 32 46,1 503 472 2021-03-17 16:00:00 5 56 22 95 33 30 521 300 05:16,3 00:29,4 2,75 28 9,69 41,9 21,5 37,6 391 268 02:03,7 00:25,9 5,05 29,3 11,64 46,2 26,1 39,8 398 268 00:59,6 00:18,8 4,51 29,3 11,57 46 26,5 40,5 394 367 2021-03-17 17:00:00 3 22 10 43 16 18 301 300 05:19,1 00:29,6 1,84 20,1 6,99 26,7 15,69 32,01 250 167 02:03,8 00:25,7 3,31 20,6 8,08 29,2 18,6 33,5 254 167 00:59,5 00:19,4 3 20,6 8,36 29,6 18,8 33,08 252 230 2021-03-17 18:00:00 1 3 2 6 3 4 59 150 05:17,1 0,42 7,2 3,4 5,79 7,3 23,4 56 35,3 02:03,8 0,74 7,2 3,4 6,4 7,3 23,4 56 35,3 00:58,3 0,67 7,2 3,4 6,42 7,3 23,4 56 48,8 2021-03-17 19:00:00 0 0 0 0 0 0 0 150 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 20:00:00 0 0 0 0 0 0 0 50 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 21:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 22:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-17 23:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2021-03-18 0:00:00 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 average 05:12,8 00:29,9 02:03,8 00:26,0 00:59,7 00:19,8
