Entry Name: "CMU-Sinda-MC2"
VAST
Challenge 2016
Mini-Challenge 2
Team Members:
Matthew Sinda,
Central Michigan University, sinda1m@cmich.edu PRIMARY
Ian Turk,
University of Wisconsin Eau Claire,turkij@uwec.edu
Jun Tao, University
of Norte Dame, jun.tao.12@nd.edu
Dr. Chaoli Wang,
University of Norte Dame, chaoli.wang@nd.edu
Dr. Qi Liao,
Central Michigan University,
Dr. Lei Shi,
Chinese Academy of Sciences, shil@ios.ac.cn
Student Team: No
Tools Used:
Javascript
D3
Excel
Approximately how many hours were spent working on this submission in total?
596 hours
May we post your submission in the Visual Analytics Benchmark Repository after VAST Challenge 2016 is complete? YES
Video
https://www.dropbox.com/s/tz0mb35cb3h4r0q/CMU-Sinda-MC2.wmv?dl=0
Questions
MC2.1 – What are the typical patterns visible in the proxy card data? What does
a typical day look like for GAStech employees?
Limit your response to no more than 6 images and 500 words.
Mobile data is delivered twice daily from the
mail delivery robot, at 9am and 2pm, and only runs during the week. An
example of morning delivery can be seen in figure 1 and afternoon delivery can
be seen in figure 2.
Figure
1. Two examples of mobile delivers during morning hours on 31 May and 1
June. Nodes represent employees logged.
Figure 2. Two examples of mobile delivers during afternoon hours on 31 May and 1 June. Nodes represent employees logged.
Both deliveries last for approximately 1 hour with the morning delivery consisting of more data entries. The robot generally moves over the same patterns every day for both mornings and afternoons respectively. When the robot delivers, it always starts on the first floor, before moving to the second and finally the third floor. In the morning, on the first floor, the robot generally starts near office 1020 before moving to an area near proxy zone 2. On the second floor, it starts near office 2100 before moving south and continuing in a counter-clockwise fashion. Once it reaches its start point, it retraces its steps and delivers to some of the offices in the center of the building. On the third floor, it starts in the vicinity of the server room before moving near office 3320 and continuing north near office 3000. Once it reaches this point, it moves back in a counter-clockwise fashion around the outside of the building, staying outside of proxy zone 5.
In the afternoon, there are significantly less data entries for mail deliver but the robot seems to follow generally the same path. In all cases, deliver seems to take less than a minute and the robot has an approximate length of stay on each floor as follows: for morning delivers – floor 1: 6 minutes; floor 2: 24 minutes; floor 3: 30 minutes; for afternoon delivers – floor 1: less than 12 minutes; floor 2: 30 minutes; floor 3: 20 minutes.
Static data is collected in various
quantities for each employee and each department. There appears to be
three shifts with minor variations in when employees start their shift.
Those employees working in Information Technology and Facilities work three
shifts – midnight to 8am, 8am to 5pm, and 5pm to midnight. The other
departments (Administration, Executive, Engineering, and HR) work one shift –
8am to 5pm. It is noteworthy that generally speaking employees do not
work on the weekends, however, there are two different anomalies to this that
will be discussed in the following sections but can be seen in figures 3 and 4.
Figure
3. Static data collected on 4
and 5 June. During this 48-hour period, only one employee, L. Carrara, is
present.
Figure 4. Static data collected on 11 and 12 June. During this 48-hour period, only three employees, M. Bramar, O. Strum, and L. Lagos have been present.
Generally speaking, employees that belong to the Engineering, Executive, and Security tend to work on the 2nd floor. Employees who work in the Administration and HR departments tend to work on the 3rd floor. The Facilities department has employees that work on all three floors, however, each employee in that department seems to have a set floor that they work on.
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.
1. All the corridors and the main entrance have the Light Power sensor at a constant nonzero value. This indicates that the lights in these areas are left on at all times.
The large meeting room, office number 1050, on the first floor and future expansion zones have the Light Power sensor at a constant value of zero, except for several short periods (less than four hours), once in the large meeting room and three times in the future expansion zone. This indicates that these rooms are rarely used or have sufficient natural lighting.
In all other zones, a regular weekday pattern
is followed in which the variable goes up to its maximum value at some point in
the day (often around 7:30am, but in some zones as late as 4pm), has occasional
dips to zero throughout the day, and returns to zero for the last time almost
exactly at midnight as seen in figure 5.
Figure 5 – Typical Light Use for the building.
In all of these zones, the variable remains at zero for all of Saturday and Sunday. The time period the lights are on in these zones indicates when the zone is in use, and a non-swing shift or night shift employee remaining in a zone with this variable at zero for more than a trivially short time could indicate malicious intent.
2. Return Outlet CO2 Concentration almost
always correlates between all zones in the buildings at any given time.
On weekdays, it follows a regular pattern of gradually increasing until a peak
at around 5pm, and then decreasing until a trough at about 11pm. This can
be observed in figure 6.
Figure 6 – Return outlet CO2 Concentration.
This indicates an estimated measure of the
total number of people in the building. On weekends, in most zones, this
variable remains constant. If on a certain day, this variable does not
follow its normal pattern, a higher-than-normal peak would indicate that more
people are in the building than usual, and a lower-than-normal peak would
indicate that fewer people are in the building. An example of which can
be seen in figure 7.
Figure 7 – Example of higher and lower-than-normal peaks of CO2 concentration.
3. In most zones, the Thermostat Temperature
sensor remains constant during the workday, including weekends, and fluctuates
near a zone-specific value at night as seen in figure 8.
Figure 8 – Example of the Thermostat Temperature sensor’s pattern for the building.
This is probably because the HVAC system is configured to save power at times when few or no people are in the building. If this sensor deviates from its normal pattern, it could indicate that the HVAC system is experiencing equipment failure or rescheduled work hours.
4. The Total Electric Demand Power sensor
follows a regular pattern, quickly rising to a high range during the workday on
weekdays and falling to a lower range at night and on weekends. This
pattern can be seen in figure 9.
Figure 9 – Example of the Total Electric Demand Power sensor’s pattern for the building.
If this variable is higher or lower than expected at a certain time, it likely means that there is an HVAC, lighting, or some other event taking an unusual amount of power.
5. The Supply Inlet Temperature sensor for
floor 1, zone 8A exhibits fairly regular peaks and valleys over the course of
the two weeks. It appears that it is most active during non-work hours as seen
in figure 10.
Figure 10 – Comparison between the Supply Inlet Temperature sensor, the Total Electric Demand Power and Dry Bulb Temperature
It is possible that in order to be energy efficient, the building uses outside night air that is already cooler and does not need as much energy. This correlates with both the peaks and valleys of the Total Electric Demand sensor and the Dry Bulb Temperature. When these two have peaks, the Supply Inlet Temperature is at a minimum value.
6. The Equipment power exhibits regular rises
and drops throughout the work week for HVAC zones with offices and floor 1 zone
1 as seen in figure 11 and
remains constant for HVAC zones with hallways. For offices, this suggests that
the equipment is powered down when employees are not present. Additionally, the
constant power usage in hallways might be security measure.
Figure 11 – Examples of power usage for zones with offices and zones with hallways only
7. Reheat Coil Power remains at zero for the majority of the data, except for some anomalous events on 7, 8, 11, and 12 June in all zones and 4 and 5 June on floor 3. This indicates that the HVAC system is usually working to cool down the building rather than heat it up. This is supported by the fact that the building dry-bulb temperature is usually greater than the thermostat temperature in most zones.
Floor 3 HVAC zone 1 consistently has a
different pattern of reheat coli power, thermostat cooling and heating set
point, thermostat temperature, and supply inlet temperature than the rest of
the building. Perhaps this indicates that the room is usually open to the
outside. These patterns can be seen in
figure 12.
Figure 12 – Examples of Reheat Coil Power to illustrate the patterns.
8. Supply side outlet temperature fluctuates
wildly during the workday and weekends from 7am to 6pm, but rises at an almost
perfectly constant rate over the night as observed in figure 13. The wild
fluctuations seem to indicate reliably when the HVAC system is active.
Figure 13 – Illustrating fluctuations in Supply Side Outlet Temperature.
9. VAV reheat damper position remains
constantly at the maximum value over the workday and weekdays. It stays low with
the occasional short dips to zero during the night. During both the day and
night on weekends, the short dips to zero continue, but the value stays at the
maximum instead of a low value when it is not dipping. This can be seen in figure 14.
Figure 14 – Example of VAV Reheat Damper Position
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.
1. 1 June, employee P. Young misplaces his
proxy card. His proxy card last transmitted at 1 June 2015 at
1:10pm. There are no additional entries for him until 2 June at 7:20am
when he has a new proxy card. It appears that he retrieves his misplaced
proxy card as his new proxy card transmits a signal in the same zone he lost
his old one as seen in figure 15.
Figure 15. Images showing P. Young’s suspicious proxy card movement
However, his new proxy card is still active and is moving separately from the old one, creating two P. Youngs in the system. The initial lost proxy card is significant because it is also at the same time as a Hazium increase in a near-by proxy zone, this will be further explained in question MC2.4.
2. There are several employees listed which are stated as working with the company, but do not generate any proxy data. As there is little information as to why there is no proxy data, it is left to wonder if all 11 are employees which do not accept the use of the proxy card system and are merely. These employees belong in either the Facilities or Security departments respectively.
3. In floor 3 HVAC zone 1, the thermostat
cooling set point exhibits strange behavior after the morning of 2 June and
continuing throughout the remainder of the two-weeks as seen in figure 16. While there are other
patterns throughout the building, none exhibit such highs and lows.
Figure 17 shows the thermostat
cooling set point for the remainder of the building’s zones as a comparison of
patterns.
Figure
16 - Thermostat cooling
set point for floor 3, HVAC zone 1
Figure 17 - Thermostat Cooling Set Point for all floors and zones
4. Additionally, over the next two weeks,
there are 6 spikes in Hazium on floor 3. It can be seen from figure 18, during these same Hazium spikes,
the VAV REHEAT Damper Position for this zone is in some open position.
This suggests that the release of the Hazium is timed with the position of the
damper position.
Figure 18 – Hazium compared to VAV Reheat Damper Position
5. The VAV Damper Position on floor 3, zone
9, exhibits interesting behavior in the evening hours of 7 June though the
afternoon of 8 June. This can be seen in figure 19. It appears that either the sensor is having trouble
communicating with the device or it is going into mechanical failure.
Additionally, while it is unknown as to how the HVAC systems are connected,
there does appear to be a correlation between Hazium in HVAC zone 1 and the VAV
Damper position in HVAC zone 9.
Figure 19 - VAV Damper Position for Floor 3 Zone 9
6. The Mechanical Ventilation Mass Flow Rate
(MVMFR) exhibits the same pattern, however, this device exhibits different
behavior on the weekend of 11 June. These values seem to double from the
previous weekend. It should be noted that there appears to be a
correlation between the MVMFR and the Return Outlet CO2 Concentration. On
the weekend of 4 June, the CO2 levels seem to drop to zero during the rapid
up/down motion of the graph during the same period. However, the weekend
of 11 June, the MVMFR values are doubled and the CO2 concentration does not
seem to be reducing. This suggests that there is a possible mechanical
failure in the system or someone maliciously altered the system to retain
poisons gases. This is possible as Hazium spikes as well. This can
be seen in figure 20.
Figure 20 – Mechanical Ventilation Mass Flow Rate compared to Return Outlet CO2 Concentration
7. On floor 2, HVAC zone 8, the thermostat
temperature sensor exhibits an interesting pattern. It happens on the evening of 8 June where
there is a large spike in the thermostat readings as seen in figure 21. While having an outside
temperature being high one day can affect the inside temperature, it is a
gradual rise, not a sudden increase. This sudden rise from 6-8 June,
appears to be out of place. It is unclear as to what caused this sudden
increase however, it could be faulty equipment or an overly hot employee who
likes the work place cool as the thermostat cooling set point also exhibits
similar patterns on the same days.
Figure 21 – Comparison between Thermostat Temp and Cooling Set Point
8. In HVAC zone 9, floor 3, the thermostat
temperature exhibits abnormal behavior on the early morning of 7 June. As seen
in figure 22, this sensor logs
an entry of going from a minimal range to a maximum value.
Figure
22 – Comparison between Thermostat Heating Set Point and Temperature
The most probable cause is someone increased the setting for the thermostat. It should be noted this zone also houses the server room. While it is possible that a server room could get hot, it is more likely that someone tampered with these sensors, possibly as a means to cause a distraction. It can be seen that prior to the event in question, the heating set point was at a constant value before dropping and sharply rising several times before returning to a constant state.
9. All zones except floor 3 zone 1 have a
higher than normal Reheat Coil Power starting on the night of Friday, 10 June
and ending the morning of Monday 13 June as seen in figure 23.
Figure
23 – Example of Reheat Coil Power
All zones in floor 3 except zone 1 have a similar event starting the night of Friday 3 June and ending the morning of Monday 6 June. During these events, the sensor has fluctuations that correlate inversely with dry bulb. This did not correspond to an increase in temperature in those zones, so something was cooling and making the HVAC system work hard. These patterns are mirrored exactly in the corresponding zone’s supply Inlet Mass Flow Rate and Supply Inlet Temperature.
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.
1. On
1 June at 1:10pm in floor 3 HVAC zone 2, the building has a Hazium event as
seen in figure 24. This
zone corresponds to proxy zone 6 on floor 3. At this same time, there is
an employee in proxy zone 3 as seen in figure 25. Proxy zone 3
corresponds to several HVAC zones one floor 3. These zones are zone 8,
10, 7, part of zone 6, part of zone 11c, 11b, and 11a. While it cannot be
determined if there is air flow between any of these HVAC zones and HVAC zone
2, it is possible that an employee could get close to proxy zone boundaries and
toss a canister with Hazium into a room. It should be noted, that the
employee in question, P. Young, has misplaced his proxy card during this same
time period and is not seen until the following day with a new proxy card. It’s
possible that during the haste of placing a Hazium container, P. Young dropped
his proxy card which would explain why there are no other data entries for him
on 1 June after 1:10pm.
Figure 24
– Hazium level for floor 3 HVAC zone 1 on 1 June
Figure
25 – P. Young’s movement prior
to the Hazium event on 1 June at 1:10pm. Note, zones with some level of
Hazium are shaded darker.
Figure 26
– Hazium level for floor 3 HVAC zone 1 on 2 June
Figure
27 – S. Flecha movement and
length of stay in each zone from 31 May through 3 June.
Figure
28 – Hazium levels for the
weekend of 11 June
Figure
29 – Return Outlet CO2 Concentration Sensor
It should be observed that as seen in figure
3, only one employee is registered as having entered the building over the
weekend. It should be noted that this employee first enters the building at 1pm
on 5 June, well after the start of the rise of CO2 levels. As the level of CO2
is an indicator of number of persons in an area, it is very probable that this
is an indication of a large group of employees present without proxy cards.
5. On floor 1 zone 2, at approximately
11:30am on 31 May, the equipment power begins a steady, but small rise and
lasting for approximately one hour as seen in figure 30.
Figure
30 – Abnormal behavior exhibited in Floor 2 Zone 2 during 31May
During the remainder of the two weeks, this
sensor is at a zero value. This is in
contrast to the Deli Fan sensor which shows multiple events over the two-week
window. The Light Power sensor initially
correlates strongly with the Equipment Power sensor, but has several additional
entries for short periods later in the two weeks. The strange part of this event is the number
of persons that are in this zone with the lights off. This can be observed in figure 31.
Figure
31 – Employee traffic patterns for 31 May from midnight to 12:30 pm.
However, it appears from figure 31 that
employees do not stay long in this zone.
This can be seen in the Length of Stay histogram on the bottom of the
figure. In this view, the length of stay
is minimal which is why it seems to not appear on the graph. It is possible that the Deli is closed which
would explain the minimal equipment power usage.