A loaded-cost review of the legacy hydraulic study loop, why a five-engineer planning team quietly burns $750K to $1M a year on model maintenance, and what it takes to compress the loop without replacing the simulator your team already trusts.
There is a number that no gas utility VP of Engineering puts on a slide.
The average hydraulic study at a North American gas distribution utility costs $15,000-plus in loaded engineering time and takes more than two weeks to build.
A five-person planning team running the standard loop burns $750K to $1M a year on model maintenance. Not on network advancement. Not on the next phase of the system. On rebuilding the model of the system as it exists today.
That math is not a vendor problem. It is a workflow problem. And it has not changed in 30 years.
The 1996 Loop, Walked Step By Step
DNV Synergi Gas shipped in 1996. Stoner Pipeline Simulator was already a decade old by then. Aspen HYSYS, SLB OLGA, and the rest of the workhorse modeling stack were built for engineers running scenarios on Windows NT workstations against networks that fit comfortably in memory. The tools have been updated several times. The loop they sit inside has not.
Here is the loop most gas utilities still run for a typical hydraulic study:
- GIS export. A planner pulls the network from ArcGIS or the utility GIS. One week, including the back-and-forth with the GIS team about what scope to include and how to handle the boundary conditions.
- Clean and reconcile. The export does not align with the simulator's data model. Pipe diameters are inconsistent. Elevations are missing on some segments. Materials are coded one way in GIS and another in the simulator. Two to three weeks.
- Hand-build the network in the modeling tool. For a full system study, three to four months. For a targeted study (a single feeder, a peak-day scenario, a regulator station retrofit), two-plus weeks even for a senior engineer who has built this loop a hundred times.
- Run the scenarios. The actual engineering work, the part the engineer was hired for, takes two weeks of focused time. Sensitivities, contingency cases, peak-day stress, MOC drafting.
- The network changed since the export. A capital project closed in. A regulator station was re-rated. A meter set was moved. The model is now memorializing a network that does not exist anymore. Back to step 1.
Every loop costs about $15,000 in loaded engineering time. A planning team running 50 to 70 studies a year burns through the seven-figure range of model maintenance without anyone in finance ever asking why.
The team you hired to design the next phase of the network is doing data archaeology on the last phase.
Why Engineering Leadership Stops Asking
This is not a competence problem. The engineers on the planning team are doing the work the way they were trained to do it, in the tool the utility licensed, on the data the GIS and SCADA teams can actually deliver. Every link in the chain is operating to standard.
The problem compounds upstream of the engineering team. By the time a hydraulic answer arrives, operations has typically already made the decision. The capital plan was committed two quarters ago. The compressor station retrofit got approved on a back-of-envelope estimate, not the calibrated peak-day model. The pressure-relief redesign went out for bid before the sensitivity sweep finished.
When the engineering answer arrives weeks late, three things happen, and none of them are good:
- The answer is filed. The decision has been made. The model becomes a memorial.
- The engineering team loses leverage in the next planning cycle. "We could not get an answer in time" gets internalized by operations as "engineering cannot move at our speed."
- The team's most experienced people start to leave. The senior engineers did not get a degree to do data reconciliation. The five-to-seven-year people who do the actual hydraulic thinking are the ones who quietly take recruiter calls.
This is the part of the model-maintenance burn that nobody puts on a slide, because it does not have a unit price. It shows up as planning team turnover, as capital projects that did not optimize for peak-day demand, and as the slow erosion of engineering authority inside the utility.
Get the WorkSync Field Ops Brief
Monthly read for upstream + midstream operations leaders. Case studies, benchmarks, and what's changing in the field. Unsubscribe anytime.
The Cost Stack, Made Explicit
Run the math directly. A loaded gas utility planning engineer in North America runs roughly $200 to $250 per hour fully loaded (salary, benefits, overhead, software seat, allocated G&A). A targeted hydraulic study runs about 60 to 80 engineer hours of pure data work before any analysis happens.
That is $15,000 of data archaeology per study. The actual hydraulic thinking, when it finally happens, is a fraction of that cost.
For a five-engineer planning team running between 50 and 70 studies a year (the typical mid-sized utility cadence), the model-maintenance line, the data archaeology line, is between $750,000 and $1,050,000 annually. That is before you count the studies that should have been run and were not. That is before you count the capital decisions that went out without a calibrated answer.
A 50 percent reduction in this line is not a productivity initiative. It is a hiring decision the utility never has to make again, because the planning team can finally absorb the workload it has been quietly deferring.
What "Closing The Loop" Actually Means
The wrong response to this math is "buy a new simulator." Every utility has a multi-decade enterprise contract with Synergi Gas, with Stoner, with the OLGA family for transmission work. The simulator is not the bottleneck. The data pipeline into the simulator is.
The right response is the same response refineries adopted 40 years ago when they moved off manual reconciliation: build the data pipeline that keeps the model continuously aligned with the network as the network changes, and put the engineer's time on the analysis that actually requires judgment.
That pipeline has four working parts:
- An automated network builder that reads the live GIS export, the SCADA tag list, and the engineering drawings (PDF P&IDs, datasheets, MOCs) and stands up the network in the simulator's native format. Not a competing simulator. A feeder.
- A continuously aligned model state that detects when the network has changed (a new feeder closed in, a regulator station re-rated, a meter set relocated) and updates the model rather than forcing a from-scratch rebuild.
- An engineering memory that remembers what was calibrated, what was assumed, what was approved, and what is currently open as an MOC. The next engineer does not start from a blank page.
- An AI engineer agent that runs the next twelve scenarios while the human engineer is reviewing the first one, drafts the MOC language from the calibrated case, and cites the source on every answer back to the drawing or the code clause.
FlowSync is built as exactly this pipeline. Model Builder reads your GIS and SCADA and stands up the network in the same week, not the same fiscal year, on 25,000-plus-node systems that solve on open. Flow Simulator runs against the network in your existing Synergi, Stoner, OLGA, or HYSYS environment. Process Simulator handles the facility cases. And Taylor, the named AI engineer agent, sits inside the loop reading every drawing in the project folder and citing every answer back to its source.
This is documented at scale today. A western US gas distribution utility runs FlowSync against more than 3.5 million meters of coverage. Study turnaround on the deployed network has compressed from a typical 200 engineer-hours per study to about 20 minutes of human review time on the agent-prepared case.
Where Taylor Fits: Apprentice, Not Replacement
A common objection from senior engineers is the right one: "I am not handing my model to an agent." Good. Neither would we.
Taylor is positioned as an engineering apprentice, not a replacement. The pattern is the same one that worked on the operations side with Willie, the WorkSync field operator agent. The engineer reviews scenario one. Taylor is already running the next twelve. The engineer reviews the citations. Taylor has already drafted the MOC language with every operating procedure and every datasheet cited inline.
The engineer who used to wait six weeks for a peak-day answer now finishes the same scenario before lunch. The senior engineer who was about to leave because they were doing data reconciliation full-time is now doing the analysis they were hired for.
This is the difference between an AI agent that competes with the engineer and an AI agent that defends the engineer's time. Utility planning teams that adopt the second pattern keep their senior people. Teams that adopt the first do not.
The 4-Week LAND Offer: $15K, You Pick The Metric
The fastest way to move a planning team off the 1996 loop is not a 12-month enterprise procurement cycle. It is a 4-week pilot on a single feeder or a single targeted study, with the success metric written down in week zero and the walk-away clause written in the contract.
WorkSync structures this as the FlowSync LAND offer: $15K, four weeks, on your live model. You pick the metric in week zero. Study turnaround. Engineer hours per scenario. Time-to-first-answer for an operations request. The pilot runs against your environment, in your simulator, with your data. If the metric moves, you sign the annual. If it does not, you walk away. No license fee. No kill fee. No awkward sales call.
A five-engineer planning team that recovers half of its model-maintenance time pays back the LAND offer in three weeks of recovered engineering hours. The Impact Guarantee is in writing.
What This Looks Like For Your Planning Team In Q1
Week 0: Baseline the current loop. Pick the metric. Sign the LAND offer. Weeks 1-2: Model Builder reads your GIS export and SCADA tag list. The network stands up in the same week. Taylor reads the project folder. Weeks 3-4: Run a real study, the next one your operations team has queued. Measure against baseline. Decision: If the metric moved, you sign the annual. If not, you walk.
For most planning teams running the math honestly, the first study compresses by an order of magnitude. The second study compresses further because Taylor has already read the project folder. The fourth study runs in the time a single status meeting used to take.
The planning team stops being the bottleneck. Operations starts asking the questions it should have been asking all along. The senior engineers who were taking recruiter calls stop taking them.
That is what closing the loop looks like in practice. Not slideware. Not a new simulator. The data pipeline you should have had since 1996, finally built.
FAQ
Why are we still building hydraulic models the same way in 2026 as in 1996? Because the simulator vendors solved the hydraulic math problem decades ago and have spent the years since adding features rather than fixing the data pipeline. The data pipeline (GIS, SCADA, engineering drawings, MOC history) is not part of the simulator's product surface. So every utility builds it by hand, every time. FlowSync is the pipeline.
Does FlowSync replace Synergi Gas, Stoner, OLGA, or HYSYS? No. FlowSync feeds them. Your existing enterprise license stays. Your senior engineers keep working in the simulator they trust. FlowSync removes the 60 to 80 hours of data reconciliation that sits in front of every study and adds a continuously aligned model state behind it.
What is the realistic study-turnaround compression on a typical deployment? The current reference deployment (3.5M-plus meters at a western US gas distribution utility) runs about 200 engineer-hours to 20 minutes of human review time, with Taylor running the agent-prepared scenarios in parallel. Most utilities see meaningful compression on the first study and an order-of-magnitude compression by the fourth.
Is the cost of model maintenance really $750K to $1M a year for a five-engineer team? Run the math on your own team. Loaded engineering cost per hour times engineer-hours per study times studies per year. Most planning leads find the number is higher than they expected, because the "studies per year" figure usually undercounts the targeted feeder studies and the engineering hours per MOC.
How is Taylor different from a generic copilot? Taylor is scoped to the engineering project folder, not the open internet. Every answer cites the source: the drawing, the datasheet, the operating procedure, the code clause. Taylor drafts MOC language in the format the utility already uses. Taylor remembers what was calibrated, what was assumed, and what is open. A generic copilot does none of these things, and a senior engineer should not use one for any of them.
What does the 4-week LAND offer actually include? $15K. Four weeks. Live model. You pick the metric in week zero. If the metric moves, you sign the annual. If not, you walk away. No license fee. No kill fee. The Impact Guarantee is in writing.
Who is the LAND offer for at a gas utility? The VP Engineering or the planning team lead. The buyer who feels the model-maintenance burn first is the engineering leader who is losing senior people to recruiter calls. Operations is the secondary buyer; they win because the answer arrives before the decision goes out.
