Module 12
Network Inventory, GIS & the Digital Twin
A pipeline operator has to know two different things about every foot of its network: where the pipe physically lies in the ground, and what that pipe is connected to. Those are two separate models of the same system — and keeping them in sync is the deepest data-modeling idea in this whole domain.
What you'll be able to do
- Explain the spatial model — GIS route geometry, alignment sheets, stationing, and depth of cover.
- Explain the logical model — the topological asset graph that answers "what feeds what."
- Articulate the spatial-vs-logical duality: why both are needed and why they must stay in sync.
- Describe a digital twin as a fusion of asset records, GIS, inspection data, and live SCADA.
- Name the KPIs operators track — throughput, leaks per mile, and LUAF among them.
Two models of one network, fused into a twin, measured by KPIs.
The spatial model — where the pipe physically is
The spatial model is the network as it sits on a map. Its home is a GIS (Geographic Information System) — software that stores the real route geometry, so you can see exactly where the steel runs across the landscape.
Field crews don't carry a GIS; they carry an alignment sheet — an engineering drawing of the route. It shows where valves sit, where the line crosses roads, rail, rivers, and other buried utilities, and how deep the pipe is buried.
🛣️ Analogy: the pipeline's odometer
A car's odometer measures distance traveled along the road, not your GPS latitude/longitude. A pipeline uses the same idea: stationing measures linear distance along the pipe itself — the line's mile-marker. "Station 142+50" means 14,250 ft down the line, no matter which way it curves on the map.
Linear referencing — locating a 1-D pipe on a 2-D map
A pipe is essentially a one-dimensional object (a long thread) drawn on a two-dimensional map. So assets are located by their measure along the line — their station — rather than by raw coordinates. This technique is called linear referencing.
Industry tooling for this includes Esri's APR ArcGIS Pipeline Referencing and the UPDM Utility & Pipeline Data Model — named here only as common industry references, not as products to buy.
An alignment sheet reads left to right by station, showing depth of cover, crossings, and features located by their measure along the line.
Depth of cover — how deep, and why it varies
Depth of cover is how much soil sits on top of the pipe. It's regulated (49 CFR 192.327 for gas) because deeper pipe is harder for an excavator to hit. The required minimum depends on where the line runs.
ℹ️ These are minimums, and they vary
Less cover is allowed in rock (it's hard to dig and hard for others to dig through); more is required under navigable waters, and liquid lines (49 CFR 195.248) generally require deeper cover. varies
The logical model — what feeds what
Now forget the map entirely. The logical model (also called the connectivity or topological model) is the network as a graph: pipe segments are the edges, and they join at nodes.
The interesting elements live at those nodes — valves, regulators, compressors, taps, and fittings. This model exists to answer one operational question: "if I close this valve, what loses supply?" — and it answers it without knowing a single coordinate.
🧭 The asset graph is pure connectivity
A graph doesn't care that two segments are a mile apart or stacked vertically at a crossing. It only cares which segment touches which, through which valve. That's exactly what you need to trace supply and plan an isolation.
Logical view: edges are segments, nodes are junctions, and valves are the switches that control connectivity.
Modern utility GIS implements this with rule-based network models — Esri's ArcGIS Utility Network (with subnetworks, connectivity rules, and topology validation) succeeded the older Geometric Network and vendor models. Again: named as an industry reference point, not a recommendation.
Spatial vs logical — the duality
Here is the centerpiece of the whole module. The spatial and logical models describe the same physical network, but they answer different questions — and you genuinely need both.
Spatial model
- GIS route geometry on a map
- Coordinates, stationing, depth
- Answers: WHERE do I dig?
- Crew dispatch, permitting, 811 locates
Logical model
- Topological asset graph
- Nodes, edges, valves, taps
- Answers: WHAT is connected?
- Valve isolation, supply tracing, outages
⚠️ The duality is COMPLEMENTARY, not redundant
The two models are not copies of each other — neither one is enough alone. The logical model tells you which three customers lose gas; the spatial model tells you which intersection to drive to. The hard part of pipeline software is keeping them in sync: when a valve moves in the field, both models must update together, or a trace will lie to you.
SPATIAL ─── tells you ──► WHERE to dig (coordinates, station, depth) │ ▲ └── same physical pipe ── must stay in sync ───┘ │ ▼ LOGICAL ─── tells you ──► WHAT is connected (graph, valves, supply)
The digital twin — fusing it all
A digital twin is a virtual replica of the live network. It fuses four kinds of data into one model so you can ask questions of the network without going to the field.
ILI = In-Line Inspection (smart-pig data); CP = Cathodic Protection; SCADA = Supervisory Control & Data Acquisition.
With those fused, the twin supports hydraulic simulation (model a flow change before you make it), predictive maintenance (spot a corroding segment before it leaks), and faster incident response (trace and isolate in software, in seconds).
At its core, that hydraulic simulation models the network's pressure gradient — the managed pressure drop from source to burner tip that is the throughline of this whole course — so the twin can predict deliverability and spot problems before they happen rather than after.
🧯 Treat vendor ROI claims skeptically
The digital-twin concept is well established and genuinely useful. But marketing numbers — "cut outages by 40%!" — are sales claims, not laws of physics. A twin is only as good as the data feeding it, and stale spatial/logical sync quietly poisons every simulation it runs.
Try it: trace & isolate
Close a valve and watch what loses supply — in both the spatial map view and the logical graph view. The two views are the same network. Toggle between them to feel the duality.
KPIs that matter
Once your data is in order, you measure the network. A KPI (Key Performance Indicator) is a number an operator watches to judge whether the system is healthy and well-run.
| KPI | What it tells you | Typical |
|---|---|---|
| Throughput | Volume / energy actually moved per period | — |
| Deliverability | Max the system could deliver under conditions | — |
| Line-pack delta | Change in gas stored in the pipe (balancing) | ± |
| Leaks per mile | Leak frequency — a standard PHMSA measure | lower = better |
| LUAF | Lost & Unaccounted-For gas (in minus metered out) | ~1–4% |
| Integrity % | Mileage assessed / compliant; dig findings | → 100% |
| MAOP / CP compliance | Operating within pressure limits; corrosion control healthy | → 100% |
📊 Reading LUAF without panic
LUAF (Lost & Unaccounted-For gas) typically runs ~1–4% (often around 2%, varies). It is a broad accounting bucket, not a single cause: compressor fuel, leaks and blowdowns, changes in line fill (line pack), and measurement error (meters at different temperatures and pressures, timing mismatches). Persistently high LUAF is a flag worth chasing. varies
MAOP = Maximum Allowable Operating Pressure; PHMSA = the US Pipeline and Hazardous Materials Safety Administration.
Key takeaways
- Every network is modeled two ways at once: spatial (where) and logical (what's connected).
- The spatial model = GIS geometry, alignment sheets, stationing (linear referencing via APR/UPDM), and depth of cover (≈30 in Class 1 / 36 in Class 2–4 & roads / 24 in distribution — varies with rock and water).
- The logical model = a topological asset graph of nodes, edges, and valves that answers "if I close this valve, what loses supply?"
- The two models are complementary and must stay in sync — that's the core data-modeling challenge.
- A digital twin fuses asset records + GIS + inspection (ILI, CP) + live SCADA; treat vendor ROI claims skeptically.
- KPIs include throughput, line-pack delta, leaks per mile (PHMSA), and LUAF ~1–4%, plus integrity/MAOP/CP compliance.
What does stationing measure?
Why: Because a pipe is a 1-D object on a 2-D map, assets are located by their measure along the line (linear referencing), not by raw coordinates.
Which question does the logical / connectivity model answer that the spatial model cannot?
Why: The asset graph traces connectivity independent of coordinates. The other three are all spatial questions — they need the map model.
Roughly what is the regulated minimum depth of cover for a Class 1 transmission line in normal soil?
Why: ≈30 in for Class 1, ≈36 in for Class 2–4 and under roads/ditches, ≈24 in for distribution mains — all minimums that vary (less in rock, more under navigable waters).
A digital twin of a pipeline is best described as a fusion of which data?
Why: The twin fuses engineering/asset records, GIS, inspection results, and live SCADA into a virtual replica for simulation and faster response — and is only as good as that data.
LUAF (Lost & Unaccounted-For gas) typically runs about how much, and what is it largely?
Why: LUAF is gas-in minus gas-metered-out, typically ~1–4% (often ~2%). Much of it comes from meters at different temperatures/pressures and timing mismatches — though persistently high LUAF is still worth chasing.