How to carry out a risk analysis for power and production

The question "should we have backup?" cannot be answered until someone has mapped which processes actually depend on power, how long they can tolerate an interruption, what it costs, and what has historically happened. Without that work, a conversation about an industrial BESS (battery energy storage system), UPS, or other resilience becomes a guessing exercise rather than a decision.

This article sets out eight concrete steps, from mapping critical processes to choosing a strategy, and ends with a checklist for preparing a dialogue with ZynexGroup or another supplier. Several of the steps are covered in depth elsewhere in this article series; here they are collected in the order a company can actually work through them.

Step 1: Map your critical processes

The starting point is not to ask whether power can disappear, but which of the company's processes truly cannot tolerate that. Beredskabsstyrelsen (the Danish Emergency Management Agency) guidance on holistic contingency planning recommends precisely this as the first step for any organisation: to map which functions are critical before addressing threats and solutions, because the rest of the analysis builds on that picture.

In practice, that means going through production lines, supporting systems, and administrative functions and asking which of them keep the company running, rather than merely being useful. The guidance recommends focusing the list on what is distinctly critical, to avoid a list so long that it becomes unusable in practice. A production line, a cold store, a control computer, or a safety function can each be critical, but for different reasons.

Step 2: Identify your critical loads

Once the critical processes are identified, the next step is to translate them into concrete electrical loads: which equipment, which switchboards, and which circuits supply exactly these processes? It is not necessarily the entire production line. A process can be critical as a whole, while only part of the equipment that drives it actually requires continuous supply for the process to continue or to shut down safely.

This translation from critical process to critical load in kW is covered in detail elsewhere in this article series, including why "the whole factory" is rarely the right design target, and how instead to identify the much smaller subset of consumption that actually carries the critical function. Typical categories across industrial environments are cooling and freezing, pumps, PLC and SCADA control, servers, process control, and safety systems, but the concrete list varies from company to company.

Step 3: Calculate downtime cost

With the critical loads in place, one can ask the question that ultimately decides whether an investment in resilience makes sense: what does it cost if power disappears at the wrong time? That is the downtime cost calculation, covered in detail elsewhere in this article series, with a model that brings together lost contribution margin, labour cost during a stop, scrap, restart cost, and delay cost into one assessment per hour.

Worth repeating here: downtime cost is not the same for all processes in the same company. One hour of interruption in a process with a short restart time and low scrap risk can cost markedly less than one hour of interruption where an entire batch is lost. Therefore a single overall downtime cost for the company is not enough. The calculation should be made for each of the critical processes from steps 1 and 2, because that is where the differences become visible and relevant for the choices that come later.

Step 4: Map restart time

Downtime cost is closely tied to a factor that is often overlooked: how long it takes from the moment power returns until the process is running normally again. Restart time is often longer than the interruption itself, because it can include cleaning, recalibration, heating of equipment, and a period of heightened quality control.

Beredskabsstyrelsen's guidance formulates precisely this as a central part of the foundation for planning: the organisation should establish acceptable downtime for each critical function, and recovery-time targets that must be shorter than the acceptable downtime, so that consequences do not become unacceptable. Applied to a production company, that means the restart time for each critical process should be known and documented, not estimated, because it both affects downtime cost and later determines how demanding any backup solution must be.

Step 5: Categorise loads by tolerance for interruption

Once critical loads, downtime cost, and restart time are mapped, the loads can be categorised by how much tolerance they actually have for interruption. Energistyrelsen (the Danish Energy Agency) guidance on emergency power for businesses uses precisely this logic with a scale from 1 second, 1 minute, 10 minutes, 1 hour, 2 hours, 24 hours to 1 week and notes that the requirement for reaction speed varies considerably from function to function within the same organisation.

In practice it is useful to gather the loads into fewer categories:

• Must never lose supply: functions where even a short voltage fluctuation or an interruption of under one second can create fault states, lost data, or a safety risk, typically sensitive control, servers, or processes with narrow tolerances.
• May lose supply for a few seconds: functions that can tolerate a short interruption while an alternative supply is connected, but where a longer interruption quickly becomes critical.
• May lose supply for a few minutes: functions where the consequence only becomes serious after some minutes, for example because temperature or pressure can be held stable in that period without active supply.
• Can be shut down in a controlled way: functions that do not require continuous supply, but where an uncontrolled interruption can damage equipment or product, and where the value lies in having time to shut down safely.
• Not critical: functions where an interruption of any duration has no noteworthy consequence for production, safety, or quality.

The categories should be adapted to the individual company, but they give a shared language for the conversation about which solution actually matches which load.

Step 6: Map historical incidents

National statistics for the Danish electricity grid show high and stable security of supply. In 2024, an average Danish electricity customer was interrupted for 26.7 minutes across 0.48 interruptions, close to the ten-year average. That national average is useful context, but it does not replace the company's own history: the risk of interruption is not evenly distributed geographically, and the statistics count only actual supply interruptions, not the other events that can in practice stop production.

It is therefore relevant to go back through the company's own operating logs, shift logs, and any SCADA history and map what has actually happened in recent years: total power outages, voltage dips, problems in electrical switchboards, unexpected machine stops, and planned interruptions from the grid company. Voltage dips are particularly worth noting, because they are often not recorded as interruptions in the official statistics, yet can still stop sensitive equipment. A technical review of voltage disturbances in industrial environments in Eastern Denmark (the DK2 bidding zone) recorded over half a year 341 voltage disturbances, where the individual measured locations experienced between 17 and 96 events each, far more often than actual supply interruptions, and those events go unrecorded unless the company itself has been tracking them. The purpose is not a precise probability of future events, but a realistic picture of what actually occurs at this particular location, how often, and when it has historically hit hardest.

Step 7: Obtain electricity data

An assessment of whether and how a technical solution makes sense requires a solid data foundation, not an assumption about how consumption "typically" looks. Danish electricity meters for larger consumers must be able to register withdrawal and delivery of electricity with a registration every 15 minutes or at shorter intervals, a data foundation detailed enough to show real power peaks and consumption patterns, not only annual totals.

The relevant dataset typically includes:

• 15-minute values for consumption: the actual consumption pattern over the day, week, and year, at a resolution fine enough to catch short, high swings.
• Power peaks: the highest, short-duration draws in kW. Why they arise, and what load they place on the installation, is covered elsewhere in this article series.
• Consumption profile: how total consumption is distributed over the day and year, and when the most energy-intensive processes run.
• Solar production, if relevant: hour-resolved production data, not only an annual production figure, because what matters is how well production matches consumption hour by hour. Energistyrelsen guidance on matching renewable energy plant to the company's process consumption emphasises that the profile for production and consumption should be stated in hourly values, because a solar plant's production and a company's consumption do not necessarily follow each other, even if the annual totals appear to match.
• Switchboard capacity: how much power the relevant electrical switchboards can carry, and how close the existing load sits to that limit.
• Grid connection: the size of the company's grid connection in amperes, a fundamental precondition for which solutions are technically and economically realistic at all.

Step 8: Choose a strategy

With the preceding seven steps in place, the company can consider which overall strategy actually matches its risk picture. The options cover a wide range:

• No backup: relevant if downtime cost is low, critical loads are few or non-existent, and the incident frequency gives no cause for concern.
• UPS for IT and control: a possible solution where the need is primarily fast, short-duration protection of servers, PLCs, or other sensitive control systems, often in minutes rather than hours.
• BESS in the standard model: relevant where the company primarily wants an industrial BESS system's commercial value in the market, without reserving capacity for emergency power, because the consequence of an interruption is limited.
• BESS with peak shaving: relevant where the consumption pattern has clear, recurring power peaks that can form part of the commercial optimisation, covered elsewhere in this article series.
• BESS with reserved capacity: relevant where the analysis shows a downtime cost that makes it worth giving up some of the commercial optimisation in order to supply selected critical loads during a grid outage, a trade-off covered in detail elsewhere in the series.
• BESS and STS: relevant where critical loads require that the transition between primary and alternative supply happens without a noticeable interruption, typically because the equipment is very sensitive to short voltage fluctuations.
• BESS and generator: relevant where the analysis points to a requirement for long-duration backup beyond what battery capacity alone can carry, a trade-off that always requires its own technical and economic design.

None of the directions is right or wrong in itself. They are answers to different risk pictures, and it is the analysis in the preceding steps that determines which one actually fits the company's processes, critical loads, and downtime cost.

The limits of the analysis

The eight steps give a solid basis for decisions, but they do not replace the technical design that a solution such as STS, reserved capacity, or other backup always requires, regardless of how clearly the analysis points in a particular direction. The analysis shows what is relevant to investigate further. It does not establish how the solution should concretely be sized or integrated into existing switchboards, or what it will cost. The analysis is also not a one-off exercise: critical processes, loads, and consumption patterns change when production develops or the company grows, and Beredskabsstyrelsen's guidance emphasises precisely that the mapping of critical functions should be updated when the organisation's circumstances change materially.

ZynexGroup's approach to system assessment

We do not take a position on whether an industrial BESS system is relevant before we have seen data. If a company comes to us with its own mapping of critical processes, loads, downtime cost, and historical incidents, together with its electricity data, we can assess far more quickly whether there is a match, and which part of the solution space is relevant: the standard model's commercial optimisation, peak shaving, or reserved capacity.

If the analysis points toward add-ons such as emergency power, STS, or backup, we coordinate the process and work with partners on the technical design and the installation itself. We do not recommend generator or UPS as a solution we ourselves supply, but a completed risk analysis makes it clear whether that type of solution matches the company's needs, independently of who delivers it.

Overall perspective

A risk analysis for power and production is a practical path from a general concern about power interruptions to a concrete, documented basis for choosing a strategy that matches the company's actual risk. The eight steps each build on the preceding one, and if a step is skipped, one loses the precision that makes it possible to distinguish between a company where no backup makes good sense, and a company where the consequence of an interruption is serious enough to justify an investment in resilience. The point is not that every company should end with the same answer, but that the answer should come from the analysis, not from an assumption that battery means backup, or that no backup is needed because the Danish electricity grid is statistically very stable.

Checklist before the meeting with ZynexGroup

The following checklist gathers what is useful to have ready before a company opens an initial dialogue with ZynexGroup or another supplier about production security:

• A list of the company's critical processes, with a short rationale for why each of them is critical.
• An overview of the concrete loads, in kW, that supply the critical processes.
• A calculated downtime cost per hour for the most important critical processes, not only for the company as a whole.
• Documented restart times for the same processes.
• A categorisation of loads by tolerance for interruption, from seconds to weeks.
• An overview of historical incidents in recent years: power outages, voltage dips, switchboard problems, machine stops, and planned interruptions.
• 15-minute metering data for consumption, preferably supplemented with solar production data if relevant.
• Information on existing switchboard capacity and the size of the grid connection in amperes.

If the company does not have answers to all points, that is not a barrier to an initial dialogue. It is often the very purpose of the first conversation to clarify which data already exist, and which need to be obtained, before a technical and economic assessment can go further. We use your data, to the extent they exist, for an initial assessment of whether there is a match between your situation and our model. You are welcome to contact us here.