Fully offline
Run the analysis on a standalone machine with exported data. No internet connection needed.
How it works
Plans that hold up, no matter what happens.
Coherence plans your plant's operations and energy investments against millions of possible futures. One forecast is never enough. Your plan stays sound when prices spike, markets shift or the process surprises you.
Tomorrow's power price: possible futures
IllustrativeMillionsFutures tested
1Plan that works
0Bets on a forecast
01 / Risk-aware optimization
Optimal on paper isn't good enough.
Most energy plans use one forecast. Reality differs, so the savings shrink or turn into losses. We plan for the full range of outcomes. We protect you from the worst ones.
Built for surprises. We give up a little of the best case. In return, you avoid the worst outcomes.
One method for operations and investments. Day-ahead schedules and ten-year equipment decisions use the same method.
You set the risk level. Choose how cautious the plan is. See what each choice costs or saves.
Annual savings across futures
Illustrative02 / Millions of scenarios
We test your plant against millions of futures before you commit to one.
Prices, weather, water or product demand, tariffs and equipment availability all vary. We combine them into millions of scenarios. A digital twin of your plant runs each one. A traditional study needs more time than that to set up one spreadsheet.
Millionsof scenarios scanned for every decision
1,000sof operating and investment configurations compared
10×faster than a traditional engineering study
- 01 / Inputs
Scenarios
Millions of possible futures for prices, weather, demand, tariffs and equipment availability.
- 02 / Physics
Digital twin
Your equipment, process physics and operating limits, tuned to your SCADA or vendor data.
- 03 / Decision
Risk-aware optimizer
Finds the plan that works well in all futures. It gives extra weight to the bad ones.
- 04 / Output
Robust plan
Schedules, setpoints and investments, with a clear range of outcomes you can show to finance.
03 / Security
Your data stays in your plant. Your people stay in charge.
Coherence runs fully offline or fully on-premises on your IT/business network. Nothing goes to the cloud. No change happens without a person's approval.
On-prem, on your network
Install Coherence inside your IT/business network, behind your firewall. Your SCADA and process data stay on your servers.
Human in the loop, always
Coherence recommends. Your operators review, approve and act.
04 / Three kinds of uncertainty
What a forecast can't tell you, and how we plan for it anyway.
Plants face three kinds of uncertainty. Each one breaks a single-forecast plan in a different way. Each one needs its own answer.
Worked example
A seawater reverse-osmosis desalination plant with several RO trains, a product-water storage tank, and a time-of-use tariff with demand charges. Power is one of its largest operating costs. (Hypothetical plant, used for illustration.)
-
01 / Short-term
Price spikes and disruptions
Some events are hard to forecast. Some cannot be forecast with the data you have.
In the example
A heat wave sends real-time power prices to ten times normal for three hours on a weekday afternoon. A week later, a high-pressure pump trips during the evening ramp. No forecast shows either event. A plan tuned to the forecast was filling the product tank that afternoon. It had every train running at the worst moment.
How Coherence plans for it
- Keep tank level and spare capacity in reserve. Size the reserve to how likely and how costly spikes and trips are across millions of scenarios.
- Plan the response ahead: which trains drop first, for how long, and how delivery still holds.
- Turn price spikes into savings, not bill shocks.
-
02 / Long-term
Market trends and recurring value
A new process line or a battery depends on assumptions about the market. It also depends on how much value repeats each year.
In the example
The plant is weighing a battery to cut demand charges. The battery's value depends on four things: the next rate case, whether demand-response programs keep paying, how much new solar flattens midday prices, and how fast water demand grows. A battery sized for last year's tariff can look good on paper and be stranded in three years.
How Coherence plans for it
- Compare every option across thousands of market futures. Options include battery sizes, an extra RO train and doing nothing. Each option includes how the plant would run around it.
- Pick the option whose payback holds in most futures, not only the expected one. This is often a smaller battery with flexible operation.
- Show the full range of returns. Your capital request can then answer finance's hardest questions.
-
03 / Execution
Running a plant is an art
A stable plant with careful chemistry can still behave differently than you expect.
In the example
Dropping from five RO trains to three at 4 pm changes feed pressure, recovery, and how antiscalant and pretreatment chemistry behave. Cycling membranes too often can speed up scaling and fouling. Pretreatment can take hours to settle. A schedule that looks fine in a spreadsheet can push a water-quality parameter toward its limit.
How Coherence plans for it
- Model ramp rates, minimum run times and start/stop limits.
- Keep a margin from chemistry and quality limits. Size the margin to how uncertain the plant's response is.
- Learn from your SCADA data how the plant responds. Tighten margins only as confidence grows. Your operators stay in the loop.
05 / Built on research
Developed over decades at Stanford and the U.S. Department of Energy.
The methods behind Coherence come from years of research at Stanford (including the WE3 Lab) and U.S. DOE national laboratories. The research covers energy flexibility, water treatment and decisions under uncertainty. We started a company to put that work on the plant floor.
15+Peer-reviewed papers
4Successful pilots
4Industries where our technology is demonstrated
- Foundations
Optimization under uncertainty
Stanford and U.S. DOE research on planning energy systems when prices, demand and operations are uncertain.
- Applied
Energy and water systems
Flexibility models for desalination, water, wastewater and industrial processes, published in peer-reviewed journals.
- Demonstrated
Pilots with California facilities
Four successful pilots, including a co-pilot run on a real-time SCADA data feed.
- 2026
Coherence Power
The research, now a product for plant managers and their teams.
06 / What you get
Decisions you can defend to your operators and your CFO.
-
Operating plan
Day-ahead schedules and setpoints for your operators to review. Each move comes with its reason.
-
Investment case
Right-sized storage, equipment and process changes. Payback and IRR show a range across futures, not one number.
-
Risk profile
How the plan performs in bad scenarios. What it costs to be more or less cautious.
-
A living model
The model re-optimizes when equipment, production or rates change. The plan stays current.
Join the wait list
See how your plant holds up across a million futures.
Send your billing and interval data. We will show you where the savings are and how well they hold up. No cost if we find none.