The National Energy Administration's explanation of the action plan to optimise power-system flexibility for 2025-2027 treats resource development, dispatch arrangements and market mechanisms as connected tasks.[1] The IEA's grid research likewise emphasises infrastructure upgrades alongside planning and management capability.[2] These materials support a system-level understanding of flexibility. They do not establish that any technology offers the same commercial return in every location.

ZHICE defines flexibility as the ability to change supply, demand or the timing of energy within specified constraints. Its value comes from solving a particular problem, not from an equipment category. Short-term fluctuations, intraday peaks and troughs, sustained supply-demand gaps and local network constraints call for different response speeds, durations, locations and recovery conditions. Companies should define the service first, then decide which resources to own or organise.

A common strategic bias is to infer demand from an existing product or asset. A company with a particular technology may interpret every flexibility need as an opportunity for that equipment. A more reliable starting point is a demand specification: where the problem occurs, when it arises, who is affected, how it is addressed today and which part an additional capability would improve. Without that specification, technology assessment can become little more than a comparison of technical parameters.

Companies should also distinguish system value from realisable income. A capability may support reliable operation, but whether it attracts payment depends on eligibility, procurement, dispatch and settlement arrangements. Strategy can build capability ahead of demand, provided it distinguishes investment supporting current operations from investment that preserves an option while rules or demand mature. The two need different budgets and review approaches.

Use a demand event as the basic observation unit, recording its timing, duration, affected area, required direction of change and recovery requirements. Events may come from company operating records or lawfully available system information, with data gaps and representativeness clearly stated. One typical day can explain a mechanism but cannot represent an entire year. The absence of an extreme event from the sample does not establish that it cannot occur.

Capabilities at different timescales are not freely interchangeable. A fast response may not be sustainable, while a large energy reserve may not be deliverable within the required time. Location matters as well: a resource elsewhere may not resolve a constraint at a particular node, however strong its technical performance. Appraisal should retain time, location and network conditions rather than reduce the resource to one aggregate capacity figure.

For flexible demand, the assessment should include production constraints. Whether a process can be interrupted, whether it needs compensating operation after restart, and whether adjustment affects quality or delivery all influence actual availability. Additional energy or work after a response also belongs in the assessment. Measuring only the load reduction during the event can overlook recovery costs and overstate the resource's net contribution to the business.

Demand analysis should produce a service specification that an operations team can execute, rather than a presentation describing market potential alone. At minimum, set out activation conditions, delivery boundaries, unavailability conditions, validation methods and failure arrangements. Technical teams can assess fit, commercial teams can design terms, and finance can identify costs against the same specification. Changes to it should be reviewed jointly.

Assessment can begin with existing capabilities: units suitable for retrofit, storage, production loads that can be coordinated and external resources under contract. Distinguish asset ownership, operating control, service dispatch rights and data-use rights. Owning an asset does not necessarily permit unrestricted dispatch; reading its data does not authorise direct control. These boundaries affect both the business model and the responsibilities it carries.

Compare options against a common service specification, verifying technical fit, confidence in delivery, recurring cost and difficulty of exit. A lower initial investment may entail higher coordination and maintenance costs. Better equipment performance may be constrained by available periods or permitted use cycles. All alternatives need an explainable lifecycle basis so purchase price does not become the sole decision criterion.

Combining resources can improve coverage but adds coordination costs. Resources may differ in response limits, maintenance cycles and communications dependencies. Aggregate capability should be calculated from resources that can meet the conditions simultaneously. Adding each resource's capability at its own best moment does not establish a commitment that can be delivered at any time. Backup resources and failure arrangements also carry real costs.

For resources without adequate operating history, describe a capability range and its conditions, then narrow uncertainty through controlled trials. Trials should remain within approved safety and authority boundaries, testing actual response, persistence, recovery cost and staff coordination. Supplier performance documents are inputs to due diligence, not substitutes for validation in the company's own setting and a completed operational handover.

The official explanation of the Basic Rules for Electricity Market Operations describes electricity-energy and ancillary-service trading arrangements and relevant participants.[3] It helps companies understand potential service categories, but local products, entry conditions, dispatch priority and settlement rules still require verification. This report does not assume that any resource necessarily earns a particular market revenue, or combine rules from different regions into an idealised business model.

Revenue assessment should begin with an executable operating schedule. Providing one service in a given period may consume power, energy or a production window reserved for another. Obligations committed earlier may restrict later choices. Adding the theoretical maximum revenue from every service ignores these conflicts and makes capital investment depend on outcomes that cannot occur simultaneously.

Cost should cover both the equipment and the business. In addition to procurement, maintenance and energy, include resource coordination, forecasting, staffing, metering, contract administration, settlement disputes and necessary backup. Business teams should confirm how production opportunity costs from demand adjustment are calculated. A service can increase revenue while reducing the company's total contribution, so appraisal must return to the net operating result.

Where prices, dispatch frequency or customer participation are uncertain, use base, constrained and improvement scenarios. The base should contain only supported revenue. The constrained case should test tolerance when resources are unavailable, dispatch is less frequent or costs rise. The improvement case should explain what additional conditions are required. All scenarios must retain the same physical constraints; genuine limits cannot be removed merely to illustrate upside.

Companies may build and operate directly, outsource operations, develop with partners or aggregate external resources. The choice depends on resource control, demand stability, technical and market experience, and willingness to bear long-term asset risk. No model is inherently superior. What matters is the rights obtained, the responsibilities accepted and whether returns correspond to the activities the company can control.

Where demand is established and core resources are controllable, operating capability can turn trial experience into an ongoing service. Where demand is clear but resources depend on external parties, prioritise authorisation, exit rights and revenue allocation. Where resources are controlled but paying demand is unproven, start with a limited pilot. When evidence is weak on both sides, additional asset investment generally cannot substitute for commercial validation.

Partnership agreements should identify who forecasts, who decides dispatch, who confirms delivery and who handles settlement disputes, along with information-sharing and exception-escalation arrangements. Even with an external operator, the company needs the ability to understand operating results and supervise critical risks. Avoid allowing all knowledge to reside with the partner. Agree data, configuration, contract and knowledge transfer at exit before the partnership begins.

Portfolio expansion must reflect the actual capacity of the operating platform. More resources make forecasting, metering, scheduling, authorisation and customer coordination more complex. Scale does not automatically reduce cost. Track staff and maintenance effort per unit of delivered service, exception backlogs and contract-change workload. Confirm the organisation can consistently support additional resources before expanding the scope of business.

A flexibility business does not earn revenue automatically once a contract is signed and equipment installed. The operations team must maintain available-resource records, committed services, maintenance plans and exception statuses, and update external commitments when capability changes. Update frequency should match service requirements, preventing outdated inventories from causing overcommitment without imposing unnecessary high-cost management on every resource.

Establish an accountability chain from resource status to delivery settlement. Resource providers confirm physical capability and constraints; operators maintain dispatch plans; on-site teams work within explicit safety boundaries; metering teams provide verifiable evidence; and finance reconciles final settlement. For exceptions, specify when commitments should be paused, who may change the plan and who explains the effects to contractual counterparties.

Delivery assessment should distinguish 'not dispatched' from 'unable to deliver'. The former may reflect demand or market conditions; the latter may arise from resources, organisation or systems. They have different implications for investment and should not be collapsed into a single revenue shortfall. Likewise, distinguish delayed response, insufficient duration, missing metering evidence and settlement disputes so each can receive the appropriate remedy.

Measures should trigger concrete action. Deliverable capability should refer to resources actually available after constraints; performance should be assessed against valid commitments or actual dispatch tasks; and settlement variance should identify the scope of completed settlement. Avoid detailed measures without reliable collection methods, and do not treat unsettled outcomes as final revenue. Definition changes should be recorded with an explanation of their effect on trends.

An effective pilot covers resource identification, contractual authorisation, operational testing, routine dispatch, settlement reconciliation and issue closure. Proving that equipment can respond once does not establish that the team can run an ongoing business or realise commercial revenue. Work backwards from the intended replicable business model to define which critical assumptions must be tested and which conditions require longer observation.

Samples should cover typical operating states and known constraints, such as resource maintenance, changing demand, partner changes and data exceptions. Live testing must respect production safety and authority boundaries. Conditions that cannot be physically tested may be assessed through a tabletop exercise, with the evidence level clearly identified. A successful exercise is not a successful field delivery; a settlement simulation is not cash revenue already earned.

Review should record both success conditions and failure conditions. A smooth delivery may depend on an experienced individual, exceptional coordination or uncounted manual effort. If these conditions cannot be reproduced consistently, scalable capability has not yet been established. Convert critical individual knowledge into rules, records and handover arrangements, and assess whether a new customer or resource requires a redesigned process.

Expansion gates can focus on evidence completeness: clear resource boundaries, effective principal contracts, validated typical states, traceable settlement and executable exception responsibilities. The aim is not to eliminate all risk, but to make the remaining risk visible to management. Even after passing the gate, increase scale in stages while observing whether organisational load and shared failures rise with the resource base.

Management can separate flexibility opportunities into optimisation of existing resources, new investment for a defined use case and longer-term exploration. The first tests improvement potential and incremental operating cost; the second verifies demand and delivery conditions; the third uses a limited budget to acquire evidence. They should share an analytical language without being held to identical revenue-certainty requirements or merged into one growth chart that hides their differences.

A practical starting point is a clearly bounded industrial park, plant or resource portfolio. Produce a demand specification, resource inventory, executable operating plan and settlement-validation plan. The project owner should be accountable for both technical and commercial validation and regularly explain unresolved assumptions to investment and business management. The pilot can then inform capital decisions instead of becoming a demonstration detached from operations.

Enduring advantage comes from continuing to understand the relationships between demand, resources and operations. Equipment can be purchased; the contract design, forecasting, coordination and review capabilities required for consistent delivery must be developed. Let verified capabilities support expansion and actual operating records refine strategy, rather than deciding asset scale first and then constructing a revenue narrative to justify it.