Entry Name: ZJU-Zhou-MC2

VAST Challenge 2016
Mini-Challenge 2

Team Members:

Zhenglei Zhou, Zhejiang University, rockstone@zju.edu.cn PRIMARY

Yubo Tao, Zhejiang University, taoyubo@cad.edu.cn SUPERVISOR

Hai Lin, Zhejiang University, lin@cad.zju.edu.cn SUPERVISOR

Student Team: YES

Tools Used:

Python

Tableau

Excel

Approximately how many hours were spent working on this submission in total?

About 220 hours on this submission.

May we post your submission in the Visual Analytics Benchmark Repository after VAST Challenge 2016 is complete? YES

Video

ZJU-Zhou-MC2.wmv

Questions

MC2.1 – What are the typical patterns visible in the prox card data? What does a typical day look like for GAStech employees?

Limit your response to no more than 6 images and 500 words.

Generally, the employees are divided into three groups considering the time they go to work and go back home. We first introduce the typical day look like for GAStech employees and then we describe typical patterns visible in the prox card data.

Typical day look like for GAStech employees:

Fig.1-1 Group 1 prox card scatter plot.

Group 1:

prox-id: vawelon001, earpa001

This group’s office hour starts from about 00:00 am to 07:00am at weekdays.

Fig.1-2 Group 2 prox card scatter plot.

Group 2:

prox-id: tsong001, agerard001, morlunv001, ibaza001, amorlun001 and other 14 people.

This group’s office hour roughly starts from 16:00pm to 23:00pm at weekdays. They tend to go downstairs to eat something or do something else at around 20:00pm and then go back to their work.

Fig.1-3 Group 3 prox card scatter plot.

Group 3:

prox-id: Other prox-id employees

The Fig.1-3 shows that this group’s office hour roughly starts from 08:00am to 17:00pm at weekdays (Except jsanjorge001, gflorez005 behave differently on July 2^nd, 7^th and 8^th). They usually eat lunch at around 12:00 pm and go back to their work at 13:00pm.

Typical patterns visible in the prox card data:

Fig.1-4 The overall analysis of working days.

P1-1:

As depicted in Fig.1-4, each blue block represents that employee go to work at the exact day. Generally, almost all employees only go to work at weekdays and have a rest at weekends.

Fig.1-5 Office 2700 during 13:00pm to 15:00pm.

P1-2:

As depicted in Fig.1-5, we notice that around 40 people of the group 2 tend to gather in 2700 (Mtg/Training office) at 14:00pm during weekdays. After this meeting, they will go back to their previous office.

Fig.1-6 Office 2365 during 13:00pm to 15:00pm.

P1-3:

Similarly to P1-6, we notice that around 10 people of the group 2 gather in 2365(Conference office) at weekdays.

P1-4:

From Fig.1-3, we notice that one employee may be off work at around 06:00pm at weekdays except on July 9^th and 10^th.

P1-5:

Another interesting finding is that the robot Rosie only travels from 09:00 am to 14:00pm at weekdays.

Additionally, it’s noteworthy that the prox-id ending with number larger than 001 seemingly refers to the same person.

MC2.2 – Describe up to ten of the most interesting patterns you observe in the building data. Describe what is notable about the pattern and explain what you can about the significance of the pattern.

Limit your response to no more than 10 images and 1000 words.

Fig.2-1 Lights power heatmap.

(1) P2-1:About Lights Power

From Fig.2-1, we notice that the lights power consumed by each floor’s corridor (F_1_Z_8A, F_1_Z_8B, F_2_Z_12A, F_2_Z_12B, F_2_Z_12C, F_3_Z_11A, F_3_Z_11B, F_3_Z_11C) and Main Entry F_1_Z_1 (inside black box) is at a constant speed every day. While other zones’ (inside red box) consuming speed change periodically at weekdays and these zones seldom consume lights power at weekends.

Fig.2-2 Equipment power heatmap.

(2) P2-2: About Equipment power

From Fig.2-2, we notice that Equipment power for zone F_3_Z_9 is much higher than other zones, and it proves that it’s a server room. After we exclude F_3_Z_9, we see a similar pattern which occurs in lights power heatmap. We conclude that the corridor and main entry consume equipment power at a constant speed. Besides, other zones show a constant pattern. Other zones’ consuming speed changes periodically at weekdays while seldom consume equipment power at weekends.

Fig.2-3 (Cooling Setting Point – Heating Setting Point) Heatmap

(3) P2-3: About (Cooling Setting Point – Heating Setting Point) heatmap

From Fig.2-3, we notice that this calculation(Cooling Setting Point – Heating Setting Point) through observation on cooling setting point, heating setting point heatmap. We find that their difference is large(around 11.1℃) at midnight, and is small(3℃) from 06:00 am – 20:00pm at weekdays. Interestingly, the difference is almost zero on July 11^th to 12^th(weekends) for all all zones except F_3_Z_1 and on July 4^th to July 5^th for the 3^rd floor except F_3_Z_1. Besides, on July 7^th and 8^th , the difference is constant at 3℃.

Fig.2-4 RETRUN OUTLET CO2 concentration Heatmap.

(4)P2-4: About RETRUN OUTLET CO2 concentration

From Fig.2-4, we notice that return outlet CO2 concentration is generally larger in F_2 than other floors while the 3^rd is smallest. The concentration is high from 07:00am to 22:00pm at weekdays. Besides, the concentration is extremely high on July 7^th and July 8^th.

Fig.2-5

(5)P2-5:

Through mutual information analysis, we find three variables AIR LOOP INLET MASS, SUPPPLY FAN OUTLET MASS FLOW RATE and SUPPLY FAN: FAN POWER have strong correlation. Then we visualize the line chart of these three variables. In Fig.2-5, we see these three variables of each floor show nearly the same tendency.

Both of these three variables have a big fluctuation and high level of value during the first weekdays’ office hour and have a flat trend at weekends while they enjoy flat trend at midnight or weekends. We also discover that the peek value increases as the floor number increases. Moreover, supply fan outlet mass flow rate is always higher than air loop inlet mass flow rate, which is guaranteed by other energy power to ensure heat to be removed. Thus, SUPPLY FAN POWER’s variation trend is in fact consistent with these two flow rate.

Fig.2-6

(6)P2-6:

Through mutual information analysis, we find water heater tank temperature has a strong correlation with supply side outlet temperature. Then we visualize these two variables’ heat map.

In Fig.2-6, we see these two variables share the same range of value and these two plots are nearly the same. Generally, the temperature in the morning is higher than other period.

Fig.2-7

(7)P2-7:

Through mutual information analysis, we find HVAC electric power has a strong correlation with TOTAL electric power. Then we visualize the line chart of these two variables.

Through observation, we find that these two variables have similar tendency. We further plot the difference between them. Interestingly, we find the difference varies periodically.

We draw a conclusion that the power consumed by other electricity have the same variation trend every weekday. The value is high during office hours and enjoy flat trend at midnight or on Sunday. Besides, the power has a small fluctuation on Saturday.

Fig.2-8

(8)P2-8:

Fig.2-8 presents the line chart of three floors’ cooling coil power. We see that three floors show the same fluctuation during weekdays’ office hours. And the peek value is reached at noon.

We also notice that the trend is flat and the value is small at weekends for the 1^st and 2^nd floor, while the 3^rd almost keeps the peek value. Another notable pattern is that as floor number increases, the value decreases, which means consumes more power to cool the floor.

Fig.2-9

(9)P2-9:

Fig.2-9 presents the heat map of three floors’ outdoor air flow fraction. In general, the fraction decreases as floor number increases, which may be another explanation to P2-8. We also notice that the fraction of the 3^rd is high during office hours at weekdays and other time is zero. While the 1^st floor and the 2^nd behave oppositely to the 3^rd floor during the first week.

(10)P2-10:

In this pattern, we conclude some constant variables and some other findings. We find Supply side Inlet Flow rate is 0.3179, Loop Temp Schedule is 60℃, pump power is 91.37 and heating coil power is 0. Besides, VAV Availability Manage Night Control Status is zero at the 3^rd floor while shows a regular trend during weekday at the 1^st floor.

MC2.3 – Describe up to ten notable anomalies or unusual events you see in the data. Describe when and where the event or anomaly occurs and describe why it is notable. If you have more than ten anomalies to report, prioritize those anomalies that are most likely to represent a danger or serious issue for building operation.

Limit your response to no more than 10 images and 1000 words.

Fig.3-1 Hazium Concentration Heatmap.

(1)3-1: About Hazium Concentration

From Fig.3-1, we notice that the F_2_Z_4 and F_2_Z_2 has similar tendency on Hazium concentration. We may guess that the zones in the same floor shows similar tendency with other zones. Referring to the problem, we know that hazium does harm to out health, so the concentration must raise out notice. Hazium is very high on the afternoon of July 11^th. Besides, the F_3_Z_1 shows high value on the morning of July 3^rd and 9^th, the F_2_Z_2 and F_2_Z_4 show high value on the afternoon of July 7^th, and the F_1_Z_8A shows high value on the morning of July 9^th . Some action may be adopted to protect employees from harm.

Fig.3-2 Thermostat Temperature for F_3_Z_1.

(2)3-2: About Equipment power

From Fig.3-2, the temperature for F_3_Z_1 is around 32℃ in the afternoon from July 2^nd to July 13^th. We think the temperature is too high for human to live in. Maybe something goes wrong with the Heating system in that zone.

Fig.3-3 Return OUTLET CO2 Concentration

(3)3-3: About Return OUTLET CO2 Concentration

From Fig.3-3, we notice that the CO2 Concentration is higher on July 7^th and July 8^th than other days. As we all know, high CO2 Concentration may lead from bad ventilation and does harm to human body. We think that the Ventilation System may goes wrong on these days.

Fig.3-4. HVAC Electric Demand Power Heatmap

(4) 3-4: About HVAC Electric Demand Power

From Fig.3-4, we notice that HVAC system mostly consume power during afternoon. While on July 7^th and July 8^th, this system only consume power at a low level. Combining with the anomalies in 3-2 and 3-3, we may say the HVAC system goes wrong these two days.

Fig.3-5

(5)3-5:

From Fig.3-5, we notice that 4 people (csolos001, lbennett001, ncalixto001 and sflecha001) is detected by robot at 14:34pm on July 10^th. Through further exploration, we find that ncalixto001 only comes to the 3^rd floor on that day and csolos001 comes to the server room at around 15:30pm on the other days. We guess that the server room may go wrong on July 10^th.

Fig.3-6

(6)3-6:

From Fig.3-6, we notice that some people(pyoung, tquiroz, morlunv, lalcaza, junger and gflorez) has prox-id ending with number bigger than 001. It’s noteworthy that they may go to work only a few days. We draw the prox card data of lalcaza and gflorez in the figure. Through further analysis, we draw a conclusion that these prox-id in fact refer to the same person.

Fig.3-7

(7)3-7:

From P2-5, we know that three variables: supply fan: fan power, air loop inlet mass flow rate and supply fan outlet mass flow rate have strong correlation. At the first weekdays, the plot enjoy big fluctuation and high value during office hours. From Fig.3-7, we notice that the supply fan power is unusual on the afternoon of July 7^th and July 8^th. These two days show something different. More precisely, the value for these two days is low and very close to zero. We think the heating system may go wrong on July 7^th and July 8^th.

Fig.3-8

(8)3-8:

From MC2.1, we know that group 2 usually go to work after 7:00am. We filter to select someone in group2 who comes to work early than 7:00 am. We find three unusual events:

<1>gflorez005 go to work at 0:00am on July 2^nd and go upstairs at around 7:56am.

<2>jsanjorge001 go to work at 0:00am on July 2^nd and go upstairs at around 14:00pm.

<3> jsanjorge001 go to work at 6:45am on July 8^th.

Fig.3-9

(9)3-9:

Fig.3-9 show the COOL COIL POWER of each floor. We know from P2-8 that the attribute usually arrive peak value at noon. While on July 7^th and July 8^th, the value is unusually low, which may represent the Heating System’s problem.

Fig.3-10

(10)3-10:

Fig.3-10 shows that there may be someone in group2 go back home alone at around 18:02 pm. These people behave differently from other employees in group2.

MC2.4 –– Describe up to five observed relationships between the proximity card data and building data elements. If you find a causal relationship (for example, a building event or condition leading to personnel behavior changes or personnel activity leading to building operations changes), describe your discovered cause and effect, the evidence you found to support it, and your level of confidence in your assessment of the relationship.

Limit your response to no more than 10 images and 1000 words.

Fig.4-1 The Lights power heat map

Fig.4-2 The employee prox card data at weekends

(1)4-1:The relationship between lights power and employee

From MC2.1, we know that most employees belong to group2. And group2 tend to go to work 8:00am to 17:00pm at weekdays. These comply with the phenomenon that most zones’ lights power consume at a high level during that period. For convenience, we further clarify the relationship between them by weekends’ prox card data and building data. As is show in Fig.4-2, lcarrara001 mainly stay at F_3_Z_3 proximity zone on July 5^th. We guess that the lights power consumed at that time in F_3_Z_10 energy zone shown by Fig.4-1 was caused by this event. Furthermore, other 3 red box show similar relationship.

We conclude that there is a casual relationship between lights power and employee. The cause is the employee comes in the exact zones and the effect is the lights power is consumed. The level of confidence is high.

Fig.4-3 Number of employees heatmap

Fig.4-4 Return OUTLET CO2 Concentration heatmap

(2) 4-2:The relationship between the number of employees and CO2 Concentration

From Fig.4-3 we see that F_2 has larger number of people than other 2 floors, and F_2 has the highest CO2 concentration in Fig.4-4. Similarly, F_3 has smaller number of people than other 2 floors, and F_3 has the lowest CO2 concentration in Fig.4-4. Moreover, we notice in Fig.4-3 that employees’ working hours is consistent to high level CO2 concentration period in Fig.4-4.

Thus, we draw a conclusion that the number of employees has something to do with CO2 Concentration.

Fig.4-5 The Equipment power heat map

Fig.4-6 The employee prox card data at weekends

(1)4-3:The relationship between equipment power and employee

From MC2.1, we know that most employees belong to group2. And group2 tend to go to work 8:00am to 17:00pm at weekdays. This comply with the phenomenon that most zones’ equipment power consume at a high level during that period. For convenience, we further clarify the relationship between them by weekends’ prox card data and building data. As is show in Fig.4-6, lcarrara001 mainly stay at F_3_Z_3 proximity zone on July 5^th. We guess that the equipment power consumed at that time in F_3_Z_10 energy zone shown by Fig.4-1 was caused by this event. Furthermore, other 3 red box show similar relationship.

Similar to 4-1, we conclude that there is a casual relationship between equipment power and employee. The cause is the employee comes in the exact zones and the effect is the equipment power is consumed. The level of confidence is high.

Fig.4-7 The Hazium Concentration heatmap

Fig.4-8

(4)4-4:The relationship between Hazium Concentration and employees

From Fig.4-7, we notice that the hazium concentration is very high on July 11^th. Thus, we visualize the time-based employee proximity card data on that day. And we find ostrum001 and mrramar001 stay at F3Z6 (correspond to F3Z1 energy zone) for a long time. After they go back home, the Hazium Concentration reach the peek value. Maybe Hazium Concentration has something to do with employees.