A mid-size district boiler plant in northern China is typically configured with three gas-fired hot-water boilers: two duty units and one standby, each sized for extreme cold weather. The operations team soon faces an awkward fact — for over ninety percent of the year the plant runs at part load, and the deep-freeze days that genuinely require three boilers in parallel can be counted on one hand. The trouble follows: run one boiler and output falls short, or it crawls at low load for months; run two and neither is fed enough. Gas consumption per unit of heat stays high, boilers start and stop incessantly, and corrosion in the flue box and condensing section worsens every year. Boiler plant group control optimization addresses exactly this question — how many units to run and how much load each should carry — a problem that looks simple but touches everything.

Why Load Distribution Is the Core: The Boiler Efficiency Curve Is Not Flat

The rated efficiency in a manufacturer datasheet corresponds to high-load operation near rated output. The efficiency curve of a gas-fired hot-water boiler is roughly an inverted U: efficiency peaks in the eighty-to-ninety percent load ratio range; going downward, the burner is forced to raise the excess air coefficient to stay stable and stack losses climb; below thirty percent load, efficiency falls off a cliff. Worse, low load inevitably brings low return water temperature, flue gas drops below the dew point, and acidic condensate clings to the heat exchange surfaces, accumulating corrosion year by year. In other words, for the same total load, the gas consumption difference between splitting one third each across three boilers and running one at full load plus one at twenty percent can exceed ten percentage points. The goal of group control is not merely adequate total output — it is keeping every running boiler on the plateau of its efficiency curve, and the load distribution strategy is the steering wheel.

Three Load Distribution Strategies: No Universal Answer, Only Matched Ones

Three strategies dominate engineering practice, each with clear trade-offs.

Equal distribution is the simplest: all running boilers share the same load ratio. The logic is intuitive and commissioning effort is minimal, suiting fleets of identical model, efficiency and wear. Its weakness shows when total load hovers in the critical zone between N boilers being insufficient and N plus one being too many: every running boiler is dragged into the low-load region, and efficiency plus condensation risk deteriorate together.

Master-slave distribution splits the fleet into a base-load unit and peak units: the base boiler runs in the high-efficiency zone for long periods and carries the main load, while peak units engage only at high load or when the base unit is under maintenance. This suits plants mixing large and small units or old and new boilers, maximizing the value of the most efficient machine. The cost is that standby units sit idle for years and need an active maintenance strategy to prevent the classic failure of a standby unit that refuses to start when finally called.

Efficiency-first dynamic distribution goes further: the control system decides which units to run and how much each carries based on real-time load and each boiler measured efficiency, rotating by running hours to even out wear. It demands the most from metering and control — every boiler heat output and gas consumption must be available in real time, otherwise the efficiency ranking has no basis. The table below summarizes where each fits.

Unit Staging: Hysteresis and Load Forecasting, No More Frequent Ignition

Deciding how many boilers to run fears nothing more than dithering at the threshold: load touches the lower bound of the two-boiler economic zone and the system sheds a unit; half an hour later load rises and it reignites. Every cold start of a gas boiler — purging, ramping — is pure loss, and frequent cycling fatigues burners and heat exchange surfaces. Two measures cure this. First, hysteresis control: asymmetric start-stop thresholds — for example, shed a unit only when load falls below ninety percent of a single boiler rating, add one only when it rises above one hundred ten percent — leaving a buffer band where operating conditions fluctuate naturally, plus a minimum-interval guard between consecutive starts. Second, load forecasting: use tomorrow weather forecast outdoor temperature to infer building heat demand, laying out the staging decision half a day ahead instead of being led by the nose of return water temperature. The methodology of load forecasting is covered in full in an earlier article: energy forecasting: predict the boiler plant load curve one day ahead with weather regression, and is not repeated here.

Return Water Temperature Management: Condensation Is the Invisible Killer

Many plants stare only at supply water temperature and treat return temperature as a passive outcome. For gas boilers the opposite is true — low return water temperature is the common source of both efficiency loss and acidic corrosion. Water vapor in flue gas condenses on heat exchange surfaces at a critical temperature around fifty-seven to sixty degrees, and return water held below that range for long periods keeps condensate clinging to surfaces and flues. There are three remedies. In early and late heating season, actively lower the supply temperature setpoint along the weather compensation curve — when outdoor temperature is mild there is no need to burn hot water, and return temperature rises accordingly. Avoid letting a running boiler cling on below thirty percent load, which is the worst disaster zone for return temperature. For primary-side constant-flow systems, evaluate a mixing bypass that blends high-temperature supply with part of the return to raise the boiler-side return temperature. All three interact with the distribution strategy: the fewer low-load operating hours, the steadier the return temperature.

The Data Foundation of Group Control: Meter First, Optimize Second

No matter how elegant the strategy, without data it is a castle in the air. The metering layer must do three things: account heat output per boiler individually, account gas and electricity consumption per boiler individually, and route critical protection signals by hard wiring. On the heat side, the ZYH-542 ultrasonic flow meter and heat meter measure instantaneous flow and cumulative heat output of each boiler individually — these are the ballots of an efficiency-first strategy: whichever boiler consumes less gas per unit of heat deserves more load. On the data side, the energy gateway consolidates gas meter, electricity, heat and running status of every boiler into one energy monitoring platform, generating per-unit consumption ledgers weekly so group control results are accepted on data rather than gut feeling. On the execution side, the ECC200 network controller carries the interlocks, sequential start-stop and alarm logic of boiler group control, communicating bidirectionally with the building automation system so staging decisions track end demand. On the safety side, make-up water and level signals must stay independent of the group control software: the MAC3 level switch provides hard-wired high-low level protection for boiler make-up tanks and condensate recovery tanks — however smart the group control is, low-water protection must never be left to program polling.

A Retrofit Case: From Fixed Rotation to Efficiency First

A tertiary hospital boiler plant (desensitized field experience) ran three gas-fired hot-water boilers of 2.1 megawatts each. The original logic staged single boilers in a fixed rotation order against outlet temperature: at load peaks one boiler ran at full load while a second ignited frequently to fill the gap, and at low load a single unit crawled below twenty percent for long stretches. After retrofitting a group control platform with efficiency-first distribution — ranking each boiler by gas consumption per unit of heat in real time and letting the most efficient carry base load, adding staging hysteresis and minimum intervals, and correcting the supply temperature curve against outdoor temperature — one heating season of operation showed: plant gas consumption down about nine to twelve percent year on year; boiler start-stop cycles down by roughly sixty percent with visibly longer burner service intervals; running boilers held in the fifty-to-ninety percent load ratio high-efficiency band; low return temperature episodes essentially eliminated, and flue corrosion inspection stretched from annual to biennial. The retrofit investment was mainly metering instruments and the control system, with a payback inside two heating seasons.

Common Pitfalls and an Implementation Checklist

Five pitfalls recur. First, watching only supply temperature while ignoring heat metering: without an efficiency feedback loop the strategy degenerates into a timed scheduler with extra steps. Second, capping running-hour imbalance nowhere: nominally rotating while one boiler quietly carries most hours, with wear imbalance surfacing only at overhaul. Third, placing start and stop thresholds too close: in the critical load zone the unit count flips every hour and the gas saved is paid back into start-up losses. Fourth, ignoring low-load condensation: corrosion costs invisible to short-term efficiency metrics burst out in year three. Fifth, group control isolated from building automation: end demand changes never reach the boiler room and staging decisions always lag half a beat.

The implementation checklist condenses to six steps: fit heat and gas metering on each boiler; verify each boiler actual efficiency curve rather than trusting datasheets alone; choose a distribution strategy and complete the interlock logic; add hysteresis and minimum intervals to staging thresholds; bring return water temperature into monitoring and alarms; and review the per-unit consumption ledger weekly after go-live. Finish these six, and the boiler plant is the same boiler plant — but the gas bill is no longer the same gas bill.