For a power-intensive facility, electricity is not merely a bill to negotiate once a year. Its cost can vary with consumption timing, peak demand, tariff design, market prices and, where applicable, participation in grid services. Yet shifting load is only valuable if production quality, throughput, equipment life and safety remain protected.

Commercial interest in this problem is growing. EU-Startups reports that Reykjavík-based SnerpaPower secured €3.4 million to support its European expansion and develop energy-management technology for power-intensive industries. The financing is evidence that investors and operators see industrial flexibility as a commercial category—not proof that every factory can save money or earn grid-service revenue.

The practical question is therefore narrower: does a facility have controllable electrical demand, enough operational visibility and suitable market conditions to make flexibility worth the implementation cost and production risk?

Where electricity flexibility can create value

Flexibility means changing when or how much electricity a facility consumes without breaching its operating constraints. It can involve rescheduling a batch, staggering equipment starts, adjusting a temperature setpoint within an approved range, charging storage at a different time or briefly reducing a non-critical load.

The strongest candidates tend to combine high electricity use with some temporal freedom. Examples may include cold storage with thermal inertia, industrial heating or cooling, pumping systems, electrochemical processes, electric boilers, compressors, water treatment and facilities with batteries or controllable charging. Suitability depends on the process, not the sector label.

A continuously operated line with narrow quality tolerances may have little usable flexibility even if its electricity consumption is large. Conversely, an auxiliary system may offer meaningful control because it can operate earlier, later or at a different setpoint without affecting output. Facilities facing time-varying energy prices, demand charges, capacity-related tariffs or curtailment arrangements have more possible value pools, but each must be verified against the local contract and market rules.

Energy-management software can consolidate meter, process and price data; forecast demand; identify peaks; recommend schedules; or issue approved control commands. Better visibility alone can reveal avoidable simultaneous loads or unexplained baseload consumption. Automated response may add value, but it also raises integration, governance and cybersecurity requirements.

Map loads before evaluating software

The first task is not choosing a platform. It is creating a load map that connects electricity consumption to production reality. Operators should classify significant loads into four practical groups:

• Non-negotiable loads: equipment that cannot be interrupted or shifted without creating safety, quality, compliance or major throughput consequences.
• Time-shiftable loads: processes that must run for a defined duration but can start within an operating window.
• Modulatable loads: equipment whose consumption can rise or fall within validated limits, such as variable-speed drives or temperature-controlled systems.
• Buffer-enabled loads: processes supported by thermal storage, material inventory, tanks, batteries or another buffer that separates energy consumption from immediate production demand.

For each asset, document rated and typical power, minimum run time, ramp rate, start-up cost, recovery time, dependencies, maintenance implications and permissible interruption frequency. Record production constraints explicitly: temperature bands, pressure limits, batch deadlines, hygiene requirements, contractual delivery commitments and safe operating envelopes.

This exercise prevents “theoretical flexibility” from entering the business case. A load is not flexible merely because a controller can switch it. Flexibility exists only when the operation can absorb the change repeatedly and predictably.

Build the necessary data and control foundation

A facility-wide utility meter rarely provides enough detail to decide which process should respond. The implementation may require submetering at line, system or asset level, with timestamps granular enough for the intended tariff or market. Meter quality, clock synchronisation, missing-data handling and ownership of the data all matter.

The energy layer also needs operational context. Useful inputs can include production schedules, equipment status, inventory buffers, process setpoints, maintenance periods, weather data for temperature-sensitive systems, electricity prices and tariff thresholds. Historical data should cover representative production patterns rather than an unusually quiet or busy period.

Controls may connect through a supervisory control and data acquisition system, building-management system, programmable logic controllers or equipment interfaces. Before integration, clarify whether the software will only monitor, recommend actions, schedule approved assets or control equipment automatically. A staged progression is often safer: visibility first, operator recommendations second, and bounded automation only after performance has been validated.

Cybersecurity must be part of the design. Operators should examine network segmentation, authentication, encryption, vendor access, software updates, audit logs and fail-safe behaviour if connectivity disappears. Any external command path into industrial equipment needs a defined approval process. The system should default to a safe operating state, not merely the lowest-cost one.

Test the economics without assuming savings

Model each value stream separately. Avoid combining avoided peak charges, wholesale-price optimisation and possible grid-service income into one headline figure because they depend on different rules and may not be simultaneously available.

Start with the applicable electricity contract and tariff. Ask which price components vary by time, measured peak, capacity reservation or consumption profile. Determine whether the operator is directly exposed to market prices or pays a fixed retail rate. A technically flexible plant may have little immediate financial incentive under its current contract.

Grid-service participation adds further questions: Is the relevant programme open to this type and size of load? Must the facility work through an aggregator? What are the minimum bid, response-time, telemetry, availability and performance requirements? How are under-delivery and imbalance treated? Revenue should enter the model only after eligibility and commercial terms have been confirmed.

Compare potential gross value with the full cost of software subscriptions, integration engineering, submeters, communications hardware, cybersecurity review, validation, operator training, ongoing monitoring and internal staffing. Include maintenance and vendor-change costs. Most importantly, assign consequences to production disruption: lost throughput, scrap, delayed orders, additional start-ups, equipment wear and management time.

A credible baseline is essential. Compare performance against what consumption would reasonably have been under the same production volume, product mix, weather and operating conditions. A simple before-and-after comparison can mistake lower output for successful optimisation. Agree on the baseline method, adjustment variables and verification process before a pilot begins.

Set automation limits and decision ownership

Electricity optimisation crosses organisational boundaries. Energy teams may see a price signal, while production teams carry the operational risk. Maintenance understands equipment limitations, finance validates value, IT and operational-technology teams govern connectivity, and site leadership owns safety and output.

Create a control policy for every participating load. It should define permitted operating ranges, maximum event duration, recovery rules, excluded production periods and conditions that automatically block a response. Identify who can approve new assets, change constraints and suspend the system.

Override responsibility must be unambiguous. Name the role authorised to reject or stop an action, specify whether an override requires a reason code, and ensure the event is logged. Operators should never be penalised for protecting safety or product quality. Repeated overrides are useful evidence that the model, process constraints or scheduling assumptions need correction.

Automation should reflect process maturity. Stable auxiliary loads with well-understood boundaries may suit automatic control. Complex batches, unusual product runs and loads with uncertain recovery behaviour may remain advisory-only. Human oversight is not an implementation failure; it is a risk control.

A practical go, pilot or stop decision

Proceed to a limited pilot when there is a measurable load, a validated operating window, a relevant tariff or market signal, adequate metering and a low-risk way to test control. Choose one or two assets rather than attempting site-wide optimisation. Establish baseline methodology, operational guardrails and success criteria in advance, including production and equipment-health measures alongside energy outcomes.

Keep the project in a visibility-only phase when consumption is poorly understood, process data is inaccessible or tariff exposure is unclear. Better monitoring may still support budgeting, anomaly detection and contract decisions without active load shifting.

Stop or redesign the case when projected value relies primarily on unverified market access, when integration opens unacceptable cyber risk, or when minor deviations could cause disproportionate production losses. The same applies if staff cannot maintain constraints and respond to exceptions.

SnerpaPower’s reported financing and expansion plans indicate growing commercial attention to this field, but an operator’s advantage will come from disciplined implementation rather than software alone. Map physical flexibility, verify local economics, measure a defensible baseline and give production teams final authority over operating limits. Electricity becomes an operating lever only when its timing can be changed without weakening the operation it is meant to serve.