Evidence checked October 31, 2026. Reported experiments are distinguished from commercial performance; all numerical examples below are hypothetical.
1. The experiment that matters takes time
In August 2026, Utah FORGE began an extended circulation test at its underground field laboratory near Milford. DOE's August 17 account described months of circulation intended to investigate pressure and production stability, water recovery and heat extraction. Those are unusually useful subjects for an energy announcement. They identify questions that remain open instead of treating successful drilling as the final demonstration. The announced experiment is a plan in progress, not evidence that every intended day has already been completed or that its final results are known. DOE account of the extended circulation test
Enhanced geothermal systems aim to turn hot rock into a usable heat exchanger. The heat is attractive because the reservoir is not waiting for daylight or a favourable wind. But identifying heat underground and delivering dependable electricity above it are different achievements. Water must travel through useful pathways, gain enough heat, return at an adequate rate and keep doing so without unacceptable losses, costs or effects on neighbouring communities.
A productive well test can therefore be genuine progress while leaving the central commercial question unresolved. A short test establishes what happened during that test. A long operating life is a prediction that must be supported by measurements, physical models and an explanation of their limits. Neither dismissing the test as meaningless nor declaring the resource proven for decades is a defensible reading.
The evidence challenge is especially important when a project is described with a single capacity number. Is that thermal output from circulating water, gross electrical output from a generator, or net electricity after pumps and other equipment consume their share? Was the number measured continuously or calculated from a moment of favourable conditions? Does it assume a reservoir temperature that has not yet been sustained?
This article treats the underground system as the object of investigation. Its analytical framework is straightforward: follow the water, follow the heat, follow the pressure, and then follow the measurement uncertainty. The purpose is not to prescribe a drilling or stimulation procedure. Those operations require qualified specialists, site-specific designs and regulatory oversight. The purpose is to show what a reader, electricity buyer or public decision-maker should expect a credible performance claim to establish.
2. What engineers are trying to connect
The Department of Energy describes an EGS development sequence that begins with characterising a site, proceeds to creating a connected reservoir, and ends with operating a power plant while maintaining that reservoir. Its explanatory material distinguishes the hot underground circulation loop from the surface machinery that converts heat into electricity. This is a useful conceptual map, not a guarantee that a particular site can complete every stage economically. DOE explanation of enhanced geothermal systems
A reservoir in this context need not be a cavern filled with water. The useful system can be a network of fractures within rock. The geometry matters because heat transfer occurs through contact between circulating fluid and the surrounding material. Two wells can communicate hydraulically without the water exploring enough hot rock to sustain useful production for the intended life.
Imagine two underground routes between an injection well and a production well. One is short and highly conductive. The other spreads flow through a much larger region but offers greater resistance. A pressure test might show excellent communication because the short route carries water readily. That result does not establish that the broader region contributes much useful heat. The easiest hydraulic connection can become a thermal weakness if most circulating water repeatedly takes it.
Conversely, a system with considerable accessible heat can be unattractive if the pressure needed to move fluid through it consumes too much electricity or cannot be maintained within operating constraints. The project must find a workable combination of heat exchange and fluid movement. Maximising one measurement in isolation is not the same as optimising the plant.
This creates several different meanings of success. Drilling success means reaching and completing an intended interval. Hydraulic success means establishing an acceptable flow response. Thermal success means recovering heat at the required rate. Operational success means maintaining that performance through changing conditions and equipment demands. Commercial success adds cost, finance, delivery obligations and the consequences of uncertainty.
A responsible announcement should identify which meaning it has earned. If a project has only demonstrated a hydraulic connection, readers should not silently supply the missing thermal and commercial evidence. Equally, a well-characterised connection is a real technical result. The correct response is to locate it on the development sequence and ask what experiment comes next.
3. Temperature is necessary, not sufficient
A high bottomhole temperature can make a project sound almost complete. It is better understood as one condition in a chain. Useful heat extraction depends on the amount of fluid moved, the change in its thermal state, the losses between the reservoir and the plant, and the surface conversion process. A very hot but poorly connected interval can deliver less useful energy than a cooler system with sustained circulation and favourable operating characteristics.
For a simplified hypothetical calculation, suppose water flows at 50 kilograms per second and gains 100 degrees Celsius. Using an illustrative constant heat capacity of 4.2 kilojoules per kilogram per degree, the thermal transfer is 50 multiplied by 4.2 multiplied by 100, or 21,000 kilojoules per second. That is 21 megawatts of thermal power. It is not 21 megawatts of electricity. Real calculations use appropriate fluid properties and operating conditions rather than assuming a constant value everywhere.
If an illustrative surface process converted 15 percent of that thermal input to gross electricity, it would produce 3.15 megawatts before auxiliary consumption. Subtracting an assumed 0.65 megawatts for pumps and other equipment would leave 2.5 megawatts net. These are teaching numbers, not estimates for FORGE or any commercial project. Their purpose is to expose the boundaries hidden by the phrase “megawatts of geothermal output.”
The difference becomes larger when conditions change. If production temperature falls, the thermal input changes. The efficiency of the surface cycle may also change, so electrical output need not decline in exactly the same proportion as the temperature difference. If maintaining flow requires more pumping, the net result can deteriorate faster than the headline thermal measurement suggests.
A useful performance chart therefore shows simultaneous measurements rather than isolated best values. Flow, injection temperature, production temperature, pressure, gross output and auxiliary demand should refer to the same operating intervals. Pairing the highest recorded temperature with the highest recorded flow from a different day creates a synthetic operating point that may never have existed.
The practical reader's question is simple: what was delivered at the electrical meter while these underground conditions prevailed? A reservoir experiment without a power plant can still provide valuable evidence. It should report thermal and hydraulic performance honestly, leaving electrical conversion as a clearly stated model rather than an implied measurement.
4. Why circulation duration changes the claim
DOE describes FORGE as a dedicated research site whose work includes drilling, stimulation, well connectivity and flow testing. Its stated purpose is to develop reproducible methods and reduce development risk. A field laboratory is particularly valuable because it can investigate mechanisms and release information that a commercial operator might not otherwise make available. It should not be mistaken for a fleet of operating power stations. DOE FORGE overview
Duration matters because different processes reveal themselves on different clocks. Some equipment problems appear quickly. Other changes develop only after continued circulation, repeated operating changes or accumulated chemical exposure. An early stable temperature can coexist with a colder region slowly advancing through the reservoir. A short observation window may end before that change reaches the production well.
That does not mean every reservoir must be operated for its full proposed lifetime before any investment is possible. Engineering routinely makes forward predictions. The issue is whether the observations meaningfully constrain the prediction. A model fitted to a few days of data may permit many possible long-term outcomes. Additional months can exclude some of those possibilities without eliminating uncertainty altogether.
Consider two hypothetical reservoirs that produce the same temperature for the first week. In one, flow spreads across a large effective heat-transfer area. In the other, a narrow channel initially draws heat from nearby rock but begins cooling sooner. During the first week, a single production-temperature measurement might not distinguish them. Pressure responses, tracer observations and longer circulation can provide additional constraints.
Test reports should therefore publish both elapsed calendar time and actual operating time. A four-month programme containing long interruptions is not equivalent to four months of uninterrupted circulation. Interruptions themselves can be scientifically useful, but their effect on the thermal history must be included. Restart conditions may temporarily improve or alter apparent performance.
The strongest account also explains why the test ended. Completion of a planned experiment, equipment maintenance, water availability, a monitoring trigger and an unanticipated reservoir response have different implications. A test that stops early is not automatically a failure, but omitting the reason prevents an informed interpretation.
The appropriate language is proportional: demonstrated for a stated duration under stated conditions, with a forecast extending beyond that observation. The forecast becomes more credible when its assumptions and uncertainty are visible, not when the observation period disappears from the headline.
5. The water balance must close
A circulation loop invites an intuitive but potentially misleading image of the same water returning indefinitely. For reservoir accounting, the important question is how much injected fluid is recovered, how much remains in the subsurface, and how much water is needed elsewhere in the system. These quantities must be defined over a common period and with consistent measurement boundaries.
The current FORGE experiment explicitly includes water recovery among its questions. That focus provides a useful reminder: describing a circulation system does not by itself establish its water balance. The following accounting framework is an analytical explanation, not a reported loss rate from the experiment. It separates the quantities that an eventual results paper would need to establish before a reader could assess continuing supply requirements.
Suppose, hypothetically, that a test injects 1,000 units of water and recovers 930 during the reporting period. The simple difference is 70 units, or 7 percent of injected volume. Calling all 70 units permanent loss would require more evidence. Some water might be temporarily stored in the reservoir or surface equipment, and the period might exclude later returns. Calling it all recoverable would be equally unjustified.
A useful balance separates measured flows from estimated storage changes and unexplained residuals. It states meter uncertainty, density corrections where relevant, and whether the comparison uses mass or volume. Hot and cold water volumes are not interchangeable without accounting for their physical properties. A neat percentage can conceal a poor balance if the measurements do not describe comparable quantities.
Water demand also extends beyond the difference between injection and production. Treatment, cooling, maintenance and other site activities may have separate requirements. Public reporting should distinguish reservoir makeup water from total facility consumption and from withdrawals that are later returned. These measures answer different environmental and infrastructure questions.
The timing of loss matters as much as its average. A large initial requirement followed by low continuing demand creates a different supply challenge from persistent losses that grow with operation. A monthly mean can hide a short period when water needs exceed the capacity of the available supply system.
For a community evaluating a proposed project, the decisive document is not simply a promise to recycle water. It is a water balance tied to the proposed operating regime, the actual source of makeup supply, relevant permissions, treatment needs and a credible response if measured losses exceed the design assumption. That is an engineering question with local consequences, not a judgement that can be settled by a technology label.
6. Pressure has an electricity bill
The underground reservoir and the surface power plant are coupled through pumping. A favourable temperature does not erase the energy required to circulate fluid. The project needs to report the pressure and flow conditions that produce its heat output, then identify how much electricity is consumed to maintain those conditions.
A simplified hydraulic relation illustrates the issue: useful hydraulic power is pressure difference multiplied by volumetric flow. In a hypothetical system with a five-megapascal pressure difference and 0.05 cubic metres per second of flow, that product is 250,000 watts, or 0.25 megawatts. If an illustrative pump system were 70 percent efficient, electrical demand for that duty would be about 0.357 megawatts. This is not a complete plant model; it deliberately excludes other equipment and site-specific hydraulic effects.
Now suppose the flow stays constant but the required pressure difference doubles. The hydraulic power in this simplified comparison doubles too. A thermal-output chart could remain unchanged while net electrical delivery falls. Conversely, an improvement in hydraulic performance could increase net delivery even without a higher reservoir temperature.
This is why statements about injectivity or productivity need a defined reference. How much flow is obtained at what pressure, with which fluid properties and operating history? A single ratio can be useful, but it does not necessarily remain constant as the reservoir evolves. Changes in temperature, pressure distribution, deposits or flow pathways may alter the response.
Pumping demand should also be placed on the correct side of the reporting boundary. If a demonstration receives electricity from the grid for its pumps, that consumption remains part of the performance story even if a future plant is expected to supply it internally. A claim about net generation cannot omit an auxiliary load merely because it sits on a separate meter.
For electricity purchasers, the relevant comparison includes seasonal and operational variation. Surface equipment may perform differently with ambient conditions, while underground operation may be deliberately adjusted. The annual net output is the integral of those changing conditions, not the peak thermal transfer multiplied by every hour in the year.
The conclusion is not that pumping makes EGS unattractive. It is that pumping is part of the technology's measurable performance. Reporting it explicitly lets genuine improvements in reservoir engineering appear where they matter: in useful energy delivered after the system has paid its own operating demand.
7. Tracers reveal pathways, with qualifications
A tracer test can provide information that a pressure measurement alone cannot. Researchers introduce a detectable marker and observe its return through the production system. The timing and shape of the return signal help investigate how fluid travels, but interpreting that signal requires assumptions about the tracer, the rock and the flow regime.
A Cornell doctoral study published in 2017 used a small experimental well field to investigate reactive tracers and heat transport. It found that flow channeling mattered and that different tracer approaches carried different strengths and limitations. Its scale and setting should remain visible: a research result from that experiment is not a direct prediction for a deep commercial reservoir. Cornell dissertation on reactive tracers
The central interpretive distinction is between where water goes and how effectively it exchanges heat along the way. A rapid tracer arrival can indicate a fast pathway, but a complete thermal forecast also needs information about contact area, surrounding rock and the physical processes governing heat movement. Fluid travel time is an important constraint, not a universal substitute for a thermal model.
Tracer recovery also has a measurement boundary. A test that detects only part of the injected tracer must consider sampling duration, detection limits, chemical behaviour and unobserved pathways. The missing fraction cannot automatically be translated into the same fraction of permanently lost water. The tracer may interact with the environment differently from the bulk fluid.
Repeated tests can be more informative than a single celebrated result. If return curves change after extended circulation, the reservoir's effective pathways may be changing. Alternatively, the apparent change may reflect a different injection schedule or sampling method. A report should separate these possibilities rather than assuming that every difference is geological.
For a non-specialist reader, the most useful questions concern the interpretation chain. What was measured directly? Which characteristics were inferred? What alternative reservoir structures could produce a similar signal? How does the tracer interpretation agree or disagree with pressure, temperature and seismic observations?
A particularly strong study states what its tracer cannot establish. That restraint is valuable because underground imaging is incomplete. Combining several imperfect measurements can sharply reduce uncertainty, but only if the limitations of each are preserved. A colourful subsurface diagram should be understood as an interpreted model unless the underlying observations justify something more precise.
8. Heat depletion is a spatial problem
Thermal decline is often described as though a reservoir were a single tank cooling uniformly. That mental model can conceal the importance of flow distribution. A large volume of hot rock may remain underground while the routes carrying most of the water become less productive. The commercial question concerns accessible heat under the actual circulation pattern, not the total heat content of a broad geological region.
Consider a hypothetical reservoir divided into two effective flow zones. One carries 80 percent of the circulating water through 20 percent of the useful heat-transfer area. The other carries the remaining water through a much larger area. Even without assigning a precise cooling rate, the imbalance identifies a vulnerability: the heavily used pathway can dominate production-temperature decline while much of the surrounding resource remains underused.
This is an analytical illustration, not a description of a measured site. Its value is to show why a bigger mapped reservoir does not necessarily mean proportionally longer useful life. The model needs to explain the distribution of flow and heat exchange, and how well the measurements constrain that distribution.
A forecast should also distinguish a threshold from a trajectory. “Thirty-year resource life” could mean remaining above a selected temperature, maintaining a minimum net output, recovering a stated amount of heat, or remaining commercially viable under assumed prices. Those are not equivalent definitions. A reservoir might continue producing heat after it can no longer meet its original electricity contract.
Interventions complicate the comparison. A forecast may assume additional wells, changes in flow distribution or other future work. Such a plan can be legitimate, but its cost, uncertainty and timing belong in the project description. A lifetime projection that relies on future investment should not be presented as the passive endurance of the initial well pair.
The useful reporting format is a set of trajectories with assumptions. Show the expected thermal response, plausible alternatives and the conditions that would trigger a revised model. Then connect those trajectories to net electrical output and operational constraints. A single lifespan number hides precisely the information a buyer needs to assess the risk.
What makes a reservoir persuasive is not the absence of projected decline. Many engineered systems degrade or require maintenance. It is the ability to measure change, predict its consequences with reasonable accuracy and adapt within a realistic operating and financial plan.
9. Seismic monitoring measures a different outcome
EGS development must address induced seismicity as a site-specific risk, not merely as a public-relations concern. The Swiss Seismological Service explains that pre-development knowledge of the subsurface is incomplete: larger structures can be investigated, but smaller features, stresses and permeability are harder to characterise reliably before drilling and intervention. That is a reason for progressively updating the assessment, not for claiming that uncertainty makes every location equally risky. Swiss Seismological Service guidance
A seismic network has at least two distinct roles. It can help researchers understand how the subsurface responds, and it can provide information for risk management. A cloud of small detected events may be scientifically informative, but the event map is not itself a map of commercially useful fluid flow. Likewise, an absence of felt shaking is not proof that every aspect of the reservoir is behaving as intended.
The public consequence depends on more than an earthquake's magnitude. Location, depth, ground motion, local conditions and the vulnerability of nearby structures affect what people experience. Reporting only a magnitude threshold can therefore leave out the basis for judging acceptable risk. The threshold should be connected to the site-specific assessment and the applicable oversight process.
Baseline monitoring matters because changes require a reference. A project should distinguish natural background activity, detected events during operations and the uncertainty in attributing individual events. Clear attribution methods protect both the community and the operator from unsupported conclusions.
A further distinction concerns the monitoring system itself. A denser or more sensitive network may detect many more small events than an older network. An increase in the published event count can reflect better observation as well as changed seismic behaviour. Comparisons should state detection capability and cataloguing methods before treating counts as directly comparable.
Readers should expect a public explanation of how information reaches decision-makers, what independent expertise is involved and how uncertainty is communicated. This is not a request to publish operational recipes. It is a request to make the accountability chain visible. The safety significance of a measurement depends partly on whether the organisation receiving it can interpret and act on it within the approved framework.
10. A traffic light is a protocol, not a guarantee
Alberta Geological Survey published guidance on traffic-light thresholds and magnitude estimation on August 12, 2026. Its specific context is hydraulic fracturing and fluid disposal in Alberta, not a universal geothermal operating code. The document is nevertheless instructive about two general issues: thresholds need a defensible basis, and magnitude estimates need consistent methods. Transferring its numerical thresholds to an unrelated geothermal site would be an unjustified shortcut. AER/AGS Information Series 164
The traffic-light metaphor sounds simpler than the underlying decision. A coloured status depends on measurements, interpretation, uncertainty and predefined responsibilities. A protocol can be carefully designed and still face events that do not unfold exactly as anticipated. It reduces risk through disciplined decisions; it does not turn a complex underground system into a deterministic switch.
DOE's January 2012 induced-seismicity protocol presents a staged approach involving screening, communication, local monitoring, hazard and risk characterisation, and mitigation planning. It explicitly describes guidance rather than a substitute for regulatory requirements. Its age should be stated when cited, and current project obligations must be checked separately. DOE induced-seismicity protocol
For public evaluation, three questions are especially important. Who has authority to change operations? What evidence supports the thresholds and response framework? How are observations after an operating change handled? The answers should be documented before a difficult event, when incentives to reinterpret the rules may become stronger.
The reporting should also distinguish an alert from a harmful event. A conservative alert can trigger attention without implying that damage occurred. Conversely, describing all events as merely “microseismic” can obscure the need to evaluate actual ground motion and local effects. The language should follow the evidence rather than serving as reassurance or alarm by default.
Independent review can test whether the monitoring and response arrangements fit the site's geology and exposure. It should include the quality of the measurements, the handling of uncertainty and the process for updating the assessment as new information arrives. A static document that never incorporates field evidence is a weak companion to an evolving reservoir.
The right expectation is not zero uncertainty. It is a credible, transparent method for managing uncertainty, including the possibility that new evidence changes what the project can responsibly do.
11. Chemistry and equipment belong in the reservoir story
A technically successful circulation test still moves real fluid through real equipment. Its chemistry, temperature and pressure influence treatment requirements, material performance and maintenance. An underground performance claim that excludes these interactions may describe an attractive reservoir while overlooking a difficult operating system.
DOE's subsurface research programme identifies coupled thermal, hydraulic, mechanical and chemical processes as central to reservoir behaviour. It also points to monitoring needs and limitations in controlling the subsurface. This is a research agenda and description of challenges, not evidence that a particular tool has solved them at commercial scale. DOE subsurface enhancement and sustainability programme
For analytical purposes, imagine two otherwise similar tests. Both recover the same thermal power. One maintains that performance with routine treatment and stable equipment condition. The other requires frequent interruptions to address deposits or component wear. Their peak output is identical, but their annual delivery and operating cost can differ substantially.
The relevant measurements therefore include the timing of interventions and the condition of equipment, not simply a before-and-after temperature. Reports should explain whether a stable flow rate was maintained by progressively changing operating conditions. If maintaining performance requires increasing effort, the unchanged output alone can hide deterioration elsewhere.
Treatment also connects the reservoir to the water balance. A project may be able to recirculate most of its working fluid while still producing a waste stream that needs management. The amount, composition and destination of that stream belong in the facility description. “Closed loop” should never be used as a casual synonym for zero environmental interaction without a precise system boundary.
Equipment tests need representative conditions. A component that performs well in a short laboratory experiment has cleared a different hurdle from one that operates through prolonged field exposure and repeated thermal cycling. Both results are useful, but they support different claims about readiness.
This is where detailed maintenance records become evidence rather than administrative clutter. They reveal whether performance is sustainable with the staffing, spare parts and intervention schedule assumed by the project. A reservoir that looks dependable only when maintenance is excluded from the clock has not yet demonstrated the service an electricity buyer expects.
12. Replication requires more than a second location
DOE's current pilots and demonstrations page describes 21 projects spanning field-scale EGS tests and exploration drilling. The listed activities include different geological settings and development approaches. The programme's language is appropriately forward-looking: drilling, characterisation and validation are activities intended to reduce uncertainty, not completed commercial achievements merely because they appear on a project list. DOE pilots and demonstrations
Replication is the test of whether lessons travel. A successful well pair at one location can establish a method's feasibility under those conditions. It does not establish identical performance in a different stress regime, rock type, depth or temperature range. The transferable achievement may be a measurement technique or drilling practice rather than an immediately portable reservoir design.
A useful comparison separates what is standardised from what remains site-specific. Surface equipment may be repeatable while underground geometry varies. A completion method may be similar across projects while the resulting flow network differs. Calling the whole project a repeatable product can conceal where the remaining geological risk resides.
The next project also benefits from knowledge gained at the first. A fair comparison should identify learning effects without treating every improvement as proof that future costs will fall at the same rate. Faster drilling, for example, can reduce one category of expenditure while reservoir uncertainty still dominates another.
Failures and partial results are particularly valuable for replication. A demonstration that explains why a method did not meet its target can prevent others from repeating an unsuitable approach. If only successful intervals and favourable sites are reported, the public record will exaggerate the reliability of the development process.
Standard data formats help because they permit meaningful comparison. Reporting a flow rate without pressure, a temperature without location, or a test duration without interruptions leaves too much ambiguity. A shared set of definitions does not force every reservoir to be identical; it makes their differences interpretable.
The evidence of a growing industry should therefore include a distribution of outcomes. How often do projects reach the intended operating envelope? What changes are needed when they do not? How much time and investment does recovery require? Those questions are more informative than selecting the best result from each project and assembling them into an imaginary universal plant.
13. A forecast should be testable before the project is old
A reservoir model can be sophisticated and still weakly constrained. The important question is not how detailed its visualisation appears, but whether it predicts observations that were not used to fit it. A model that explains yesterday's data after extensive adjustment has passed a different test from one that accurately anticipates the next operating period.
One practical approach is to define prospective checks. Before a new test phase begins, the team records expected ranges for selected measurements under the planned conditions. The subsequent observations can then reveal whether the model is useful, overconfident or missing important mechanisms. This article proposes that approach as an accountability practice, not as a claim that every project already follows it.
Uncertainty should be propagated through to the service being sold. A range of plausible reservoir temperatures, flow responses and auxiliary loads produces a range of net output trajectories. Reporting only the central temperature forecast can conceal a much wider uncertainty in delivered electricity or cost.
The correlations matter. An unfavourable change in one variable may coincide with an unfavourable change in another. Assuming every uncertainty is independent can understate the combined downside. Conversely, simply adding all worst cases may create an implausible scenario. The model should explain how physical relationships and observed evidence inform the combinations considered.
A useful review also asks which uncertainty is reducible. More monitoring may improve knowledge of present behaviour. A longer test may constrain thermal decline. Some uncertainty about future geology or rare events may remain despite substantial effort. Investment decisions become clearer when the report identifies what another experiment could realistically resolve.
The commercial consequence should be explicit. If the project underperforms, does the owner drill another well, accept lower output, purchase replacement electricity or change the operating plan? Each option has costs and limits. A forecast that assumes effortless correction of every adverse outcome is not a serious description of risk.
The strongest performance case is therefore a sequence of predictions and checks, with revisions preserved. Changing a model in response to evidence is normal scientific practice. Quietly replacing the old forecast and presenting the new one as though uncertainty never existed is what undermines confidence.
14. What a convincing underground demonstration would show
The most useful geothermal evidence package would let an independent reader reconstruct the performance claim. It would identify the test interval, operating schedule, instrumentation, measurement uncertainty and reasons for interruptions. It would show simultaneous flow, temperature and pressure histories, together with the water balance and auxiliary energy demand relevant to the claim.
It would then connect those observations to the inferred reservoir behaviour. Tracer results, seismic monitoring and other diagnostics would be presented as complementary evidence with stated limits. The report would distinguish direct measurements from interpretation and long-term prediction. It would explain whether several different reservoir models remain consistent with the data.
For a power project, the final step would connect underground performance to net electrical delivery. The calculation would include surface conversion, operating loads, maintenance and the assumptions behind any future interventions. A thermal demonstration without a generator would remain valuable, but its electrical forecast would be labelled as a forecast.
The same package would address local consequences. Water supply and treatment, seismic risk management and the responsibilities for responding to unexpected results would be visible alongside the energy figures. These are not optional subjects to be discussed only after a technical success has been declared. They help define whether that success can become a viable project at the location in question.
The August 2026 FORGE test matters because sustained circulation is where several of these questions meet. Its eventual findings should be assessed on what was actually measured and for how long. An announced experiment cannot yet answer its own research questions. That is precisely why conducting it is worthwhile.
Enhanced geothermal does not need exaggerated certainty to deserve attention. It needs a growing body of field evidence that turns favourable geology and promising engineering into predictable service. The decisive progress will be visible in complete operating histories, improving forecasts, manageable local effects and repeated results across appropriately characterised sites.
Hot rock is the starting condition. The accomplishment is a system that can keep moving useful heat through the ground, into a plant and onto the grid, with its losses, costs and uncertainties accounted for. The most credible projects will make that chain easier to inspect, not ask the public to infer it from the temperature at the bottom of a well.