Dairies: safety margins on fat, heat and cleaning cost money

Several boilers on fixed pressure, load shared out by habit

What happens. The boiler house has two, three or more boilers. Each has its own pressure controller, and the load split is habit: the newest leads, the oldest follows. When steam demand jumps, at the start of cleaning for instance, pressure dips and every boiler reacts at once.

What it costs. Boilers running at poor part-load efficiency while another could carry the load better, burners cycling on and off, and pressure kept higher than needed so the dips stay inside limits. That is fuel spent every hour.

Why the usual fix fails. A fixed sequence or a load split set once at commissioning does not follow changing demand and boiler condition. Raising the pressure setpoint hides the dips but costs fuel and losses all day.

How a fixed-price remote diagnosis checks it. From boiler data (fuel, steam flow, pressure, flue-gas temperature and oxygen, burner states) and the production plan I check how efficiently each boiler runs at each load and when demand jumps. The fixed-price remote diagnosis shows whether an optimised load split with predictive pressure control could run with less fuel and a lower pressure margin.

Background: Economic predictive control

The end of fermentation is judged by samples

What happens. Yoghurt, quark and other fermented products are cooled when the acidity reaches its target. Someone takes a sample, measures, waits and samples again. Starter cultures, milk composition and temperature differ from tank to tank, so the time to the target is never quite the same.

What it costs. Tanks stopped too early or too late give products that are too mild or too sour, and texture that varies. Tanks waiting for the next sample block the line schedule and tie up staff.

Why the usual fix fails. A fixed fermentation time ignores how differently each tank behaves. More frequent sampling costs staff time. Inline acidity probes are possible, but they need care and still show only the current value, not when the target will be reached.

How a fixed-price remote diagnosis checks it. From temperature logs, acidity measurements and batch records I check how well the progress of each fermentation can be computed and the end predicted. The fixed-price remote diagnosis shows whether a computed fermentation value could time the cooling more precisely and what data it would need.

Fat and protein are standardised with a safety margin

What happens. Milk, cream and cheese milk are standardised to fat and protein targets. Because the inline measurement drifts and the lab result comes later, the setpoint sits above the declared minimum. Raw-milk composition changes with season, feed and supplier, and the margin covers it all.

What it costs. Every tenth of a percent above the target is milk fat or protein given away for free. Across the volume of a dairy that adds up to real money, especially when milk prices are low and margins thin.

Why the usual fix fails. Calibrating the inline analyser more often helps, but it still drifts between checks. Tightening the margin by hand is tried until one lab result is below the minimum, and then the margin comes back larger.

How a fixed-price remote diagnosis checks it. From inline readings, lab results, flows and raw-milk data I estimate how much of the margin is really needed and how well the composition can be computed between lab checks. The fixed-price remote diagnosis shows whether standardisation on a computed composition could run closer to the target, and what giveaway that would save.

Heat treatment runs hotter than safety requires

What happens. The pasteuriser or heat-treatment unit runs at a temperature and holding time well above the minimum. Nobody wants to be the one who lowered it. Flow, holding time and temperature vary, so the setpoint carries a margin for the worst moment.

What it costs. Heating and cooling energy for every litre, more fouling on the heating surfaces and therefore more frequent cleaning, and products that taste more cooked than they need to.

Why the usual fix fails. Lowering the temperature by hand is not an option without proof. The safety requirements are fixed, and rightly so; the margin above them is what nobody can quantify.

How a fixed-price remote diagnosis checks it. From temperature, flow and holding-tube data I compute the actual heat treatment each portion of product received, moment by moment, against the required minimum. The fixed-price remote diagnosis shows how much margin above the requirement you are running and whether control on the computed heat treatment could hold the requirement reliably with less excess.

Evaporator concentration drifts as the heating surfaces foul

What happens. In the evaporator, milk or whey deposits slowly coat the heating surfaces. The concentrate drifts, operators correct steam and flow by hand, and the run ends with a cleaning stop when performance falls too far.

What it costs. Concentrate outside specification, which matters for the dryer and the powder downstream, more steam or vapour recompression power, and production lost to cleaning that comes earlier than planned.

Why the usual fix fails. Fixed run times ignore how fast each run fouls. A density meter at the outlet shows the drift, not the fouling behind it, and a standard controller on it reacts late.

How a fixed-price remote diagnosis checks it. From temperatures, pressures, flows and density readings I estimate concentration and fouling through the evaporator over each run. The fixed-price remote diagnosis tells you whether a computed concentration and fouling state could hold the concentrate tighter and time the cleaning better.

Curd cutting time is judged by eye

What happens. In the cheese vat the cheesemaker decides when the coagulum is firm enough to cut, by eye and by hand. Milk composition, temperature and rennet activity change, so the right moment shifts from vat to vat.

What it costs. Cutting too early loses fat and fines into the whey; cutting too late gives a different texture and moisture. Both mean lower yield and more variation in the cheese.

Why the usual fix fails. A fixed time after renneting ignores how milk changes over the season. The experienced cheesemaker’s eye is good, but it is one person, and the judgement differs between shifts and sites.

How a fixed-price remote diagnosis checks it. From vat temperatures, milk composition, rennet dosing, cutting times and the resulting yield and moisture I check how well the coagulation state can be computed. The fixed-price remote diagnosis shows whether a computed coagulation value could support the cutting decision, and which measurements it would need.

Cleaning in place runs long, fixed cycles

What happens. Tanks, lines and heat exchangers are cleaned with fixed programmes: rinse, lye, acid, rinse, each with a set time and temperature. The programmes were written for the dirtiest case. A lightly soiled tank gets the same cycle as one after a long run.

What it costs. Water, cleaning chemicals, heat and wastewater on every cycle, and production time lost while the line is being cleaned. In a dairy, cleaning takes a large share of the day.

Why the usual fix fails. Shortening programmes by hand risks hygiene, so nobody does it. Turbidity or conductivity sensors in the return line show phases, but on their own they do not tell whether the surfaces are clean.

How a fixed-price remote diagnosis checks it. From cleaning records (flows, temperatures, conductivity, turbidity, times) and the production history before each cycle I check how much of each cycle actually removes soil. The fixed-price remote diagnosis answers whether ending each phase on the computed cleaning progress could save water, chemicals and time without risking hygiene.

Volume from closed sites lands on plants that keep running

What happens. Dairy groups close smaller sites and move their volume to the larger plants, often with new lines and new packaging formats. The remaining plants run more products, more changeovers and more cleaning on the same infrastructure.

What it costs. Line schedules get tight, cleaning cycles collide with production, and pasteurisers, fermentation tanks and cooling run closer to their limits. Energy demand peaks rise.

Why the usual fix fails. Planning by spreadsheet and experience works until the plant is full. New equipment helps but takes time, and the bottleneck often moves to utilities such as steam, cooling or cleaning.

How a fixed-price remote diagnosis checks it. From production plans, line and cleaning records and utility data I look for where the plant actually runs out of capacity. The fixed-price remote diagnosis shows whether optimised line and cleaning schedules, and better control of heating, fermentation and cooling, could carry the extra volume.

Background: Optimal paths and setpoint plans