1. The document discusses how spatial decision support and analytics can be applied at the campus scale by integrating various data sources such as GIS, CAD, BIM, and Tableau. 2. A key challenge is that a campus is a complex system with different processes and specialized data silos. The presentation explores using GIS as an enabling technology to create a comprehensive spatial model and dissolve these silos. 3. Examples of spatial decision problems on campus include optimal space assignment and indoor routing. Solutions involve building spatial databases and networks from CAD floor plans to support optimization and scenario analysis.

1. FME World Tour 2015, Boston. April 14th, 2015
Spatial Decision Support and
Analytics on a Campus Scale:
Bringing GIS, CAD, BIM and Tableau Together
Alexander Stepanov
GIS Architect
A&F Administrative Systems
Niels La Cour
Senior Planner
Campus Planning
University of Massachusetts, Amherst

3. Introduction
• Smart Campus (what is “Campus”?)
• Analytics: Descriptive, Predictive & Prescriptive (being ‘smart’
through optimization of operations and use of existing
resources).
• What are prerequisites for bringing optimization and operations
research (OR) methods to support decision-making process
• GIS as enabling technology

4. Campus Environment

5. Campus is a complex system (decomposition)
Processes
• Academic/Teaching
• Research
• Operations
• Residential Life
“No buffer space”
“Changes all the time”
“Divide & Conquer”

6. Campus as Systems: Buildings
Class Scheduling

9. ADA Pickup Locations

11. 3. Campus is a Complex System
Spatial Data: CAD, GIS, BIM, NavisWorks
Non-Spatial Data: MS SQL Server, SharePoint, Excel, Other DBs, Text files
Diff IT
Systems
• Specialization (to cope with complexity)
• Multiple databases
• “Silos” & “Silos Paradox”
• Living organization/changes!
Location/Space & Time as KEY to join and
relate entities/features in the DBs.
How can we create & maintain
‘comprehensive’ spatial model/db of
Campus efficiently.

12. 4. Data & Analytical Workflows (WF)
Client
Applications &
Systems
Desktop
Mobile
Web
3D ModelingOPTIMIZATION [ SAS/OR]
SCENARIO VISUALIZATION
ANALYTICAL TOOS
Spatial Data Warehouse
Set of ETL workflows
“PULL” system vs “PUSH System”
LEAN approach

13. Principles of such integration (CAD&GIS)
… given the fact that
majority of the
existing buildings
data are in CAD/BIM
systems/formats.

14. 2.2 3D Modeling: Tools: FME
Library of
Transformers Data Sources Data DestinationsTransformers
ETL = Extract – Transform - Load
~300 formats in
~250 formats out
Ability to embed
Python sci libraries
Easy to express ideas

15. 4. Data & Analytical Workflows (WF)
Client
Applications &
Systems
Desktop
Mobile
Web
3D ModelingOPTIMIZATION [ SAS/OR]
SCENARIO VISUALIZATION
ANALYTICAL TOOS
“BRAIN”
“HEART”
“LUNGS”

16. Decision Problems on Campus
Scale

17. Decision Problems on Campus Scale
• Within Spatial Context
• Effect ‘other’ sub-systems
• KPIs have spatial component
• Awareness of trends/projects
• Require integration of multiple data-sources
• Snapshot of performance matrix
• Snapshot of current ‘state’ of the system

18. Example: “Backfill” Space Assignment
New building is built on campus
Researchers/Groups are moved to the new building, vacating their offices and labs.
What would be an optimal policy to assign the vacant space to other groups/users?
Demand - Supply - Assignment
What parameters and constraints? Performance metrics?
Everything has a SPATIAL component

19. “Backfill” Space Assignment“Backfill” Space Assignment
Demand Supply
Cij
Xij = {0,1}

20. Creating room/space inventory

21. 3.1 Spatial DB of Space Inventory
Input:
CAD Floor Plans
CAFM
GIS Layers with BLDG
footprints
Elevation profiles
3 step process: (3 feature classes)

22. 3.1 Unique Campus Space ID & 3D representation
of rooms
0318_04_0020A
Spatial database is a collection
of rooms(spaces), which are
registered in physical space and have
unique “global” campus ID.
Connection to the enterprise DB

23. From CAD Space to ARCH Space
Origin (0,0) is shifted to specific tile/cell
STEP 1: CAD to GIS (Architectural Space)
Every SPACE
has PRIMARY KEY
BLGD_FLOOR_ROOM

24. ETL: CAD to Arch Space

25. 2D (2.5D) Inner Space Modeling: “Architectural Space”

26. 2D (2.5D) Inner Space Modeling

27. Step 2: Getting Shift, Scale and Rotation for Buildings

28. Step 2: Getting Shift, Scale and Rotation for Buildings

29. Step 3: GIS “Arch” to GIS “Real World”
Every SPACE
has PRIMARY KEY
BLGD_FLOOR_ROOM
Ability to JOIN
External datasets

30. ETL: Arch to GIS Space

31. 2-way translation: ETL to add/update data in CAD via GIS (“Breadcrumbs”)
2D (2.5D) Inner Space Modeling: 2-Way process

32. Each space/room has LOCATION_ID to join with DBs

35. Room/Space Spatial Database
3 Feature classes:
Floors (~1,800 )
Rooms (~32,000)
Arch Details (windows, doors, stairs) ~ 8 million
Every feature has LOCATION_ID

36. Buildings Networks I
(“Pine branches”)

37. • A building is more than just set of
Rooms/spaces.
• From inventory/reporting to
optimization and analysis
• Vertical & Horizontal Conveyors
Example2: Indoor Routing & Navigation
Buildings as
complex systems

38. Getting ideas for modeling building networks
IBM (Supply Chain Optimization)
Network/Graph G(N,E)
edges E:
Eh - horizontal edges (room -> floor
centroid)
Ev - vertical connections b/w floors
nodes N:
Nr - room centroids
Nf - floor centroids
G(V,E) should be connected to pedestrian
networks.

39. Testing our ideas for modeling building networks

41. Workflow to build the network/graph G(V,E)
1 2
3
4
5
5
6
Input:
FC - Rooms (~32K)
FC - Floors (~1,8K)
Output:
FC - Nodes (Pnts)
FC - Edges (Lines)

42. Network/Graph G(V,E)
|E| ~ 32,000 (horizontal) +
1,800 (vertical)
|V| ~ 32,000
Full set of attributes for
rooms/floors
Processing time: ~ 2 min.
Connectivity to
pedestrian/transportation
networks
Bridging the “scale” divide
(geo-design)Demo 2

43. Buildings Networks II

44. Building/Floor topology defines the performance metrics
(“Pine branch” networks are not enough)
IBM (Supply Chain Optimization) ESRI (Travel Time along actual road network)

45. Modeling paths between rooms
Source: J. MacGregor Smith (MIE, UMass Amherst)

46. Developing networks to model crucial building functions
Space Assignment
Location/Allocation
User/Space Interaction
2D/3D Routing
Evacuation Planning
Accessibility
Building Indoor Networks
Connection to ‘outdoor’
sidewalks/roads

47. RECTILINEAR STEINER MINIMAL TREES (RSMT) with
Obstacles
•Rooms
•Doors
•Corridors (Circulation )
•Walls ( Obstacles )
Prof. J.MacGregor Smith (jmsmith@ecs.umass.edu)
CAD/DB
Network Representation
(Floor/Building)
Example3: Indoor Routing & Navigation
Rectilinear Steiner Minimal Trees
is a graph connecting
all rooms/corridors with edges and
ensures min. total length
RSMT algorithm
(Prof. Smith J.M)
Data InterOp Extension

48. Example 3 RSMT: IMPLEMENTING MST with Custom
Transformer

49. 21 rows of code to implement MST transformer
4 rows to solve Minimum Spanning Tree sub-problem using NetworkX (LANL ) (http://networkx.lanl.gov/)
Example 3 RSMT: IMPLEMENTING MST with Custom
Transformer

50. CAD to GIS (+ text/data mining)

51. 3.2 EXTRACTING TEXT INFORMATION
SEE U702-2 (1955) LOC. APPROX.
6,000 notes
30 year of experience
Period 1940s – current
4 layers: STEAM/ELECTRIC/WATER/DRAINAGE
Non-structured text

52. 3.2 EXTRACTING TEXT INFORMATION - FME WORKFLOW

53. Regular expressions allows one to do advance search for patterns ( learn once, re-use everywhere)
We use the same approach to extract utility attributes from CAD annotation layers
3.2 EXTRACTING TEXT INFORMATION

55. 8 transformers ( + 8 transformers to parse file system for scanned documents )
3.2 EXTRACTING TEXT INFORMATION

56. Integration of Tableau (BI) and
Spatial Data on Campus Scale

57. Gartner's Magic Quadrant for Business Intelligence and Analytics Platforms
Tableau is a powerful and
an intuitive BI package.
Many examples and use
cases are based on a
regional scale.
How we can integrate it
with GIS on a Campus
scale?

58. http://www.tableau.com/solutions/higher-ed
Tableau in Higher Education

59. Default Tableau map (service) UMass-Amherst GIS map ( + 200 Buildings)
How we can approach reporting on building level with Tableau?
a) How to bring GIS Building footprints into Tableau shapes?
b) How to bring centroids of Buildings as X/Y (lat/lon) into Tableau?
Example of case use: Teaching/Instruction/Operations happen in Buildings …

60. Converts BUILDINGS footprints from GIS representation to SQL Table
which represents a line by storing shape_id, point sequence and lat/long.
Most complex part is “VertexCounter” transformer ….
Available as a free custom transformer from FME Store …
Programming time – 0.
Example of simple FME workflow to bring any polygon/line GIS geometry into Tableau

61. Building Footprints in Tableau.
Now it’s ready to be connected with data tables and to run analytics

62. FME to convert Building Footprints to centroids (Polygons -> X/Y centroid -> Lat/Lon):
Student distribution at 10am on Thursday (Fall 2013)

63. Buildings and Campus as systems: GIS as
Decision Support System
UNIQUE ANALYTICAL ABILITIES TO DECISION MAKERS
http://gis.larc.nasa.gov/spaceopt/overview.html
• Energy Use Optimization
• Facilities Management
• Asset Management (Equipment)
• Location/Allocation of Resources
• Public Safety and Security
• Maintenance
• Space and Group Management (NIH, the largest research organization)
• Optimization of Space Utilization
• Improve Collaboration
• Indoor Navigation and Routing
• Event Scheduling (‘time table for College’)
• Urban Planning
• Landscape Architecture and Design
• Sea Ports, Airports, Shopping Centers …

64. Conclusion
• GIS is an enabling technology for prescriptive analytics (Operations Research).
• GIS facilitates research collaboration among multi-disciplinary teams
• Geospatial approach dissolves silos within organization and stimulates organization-wide
awareness.
• Spatial DB/ETL coupling allows to bring existing analytical tools and
libraries to transform/analyze data.
• (Format agnostic, data centric)
• GIS approach may encourage collaboration among academic, research, operations and industry.

65. Acknowledgments
•Campus Planning, D&CM, UMass Amherst
•Prof. James MacGregor Smith, Department of Mechanical and Industrial Engineering
•Lukasz Czarniecki, Jonathan Contract, Niels La Cour and Ludmila Pavlova, Campus Planning
•Safe Software (http://www.safe.com)