Foundries: melting at shift start, long cooling, scrap found late

Induction furnaces melt at shift start, not when power is cheap

What happens. The melters charge and switch on the induction furnaces when the shift begins, because the metal must be ready for pouring. All furnaces in the foundry draw their peak power together.

What it costs. Power at the highest price hours and peak charges on the grid connection. For a foundry with high energy use and under cost pressure, this is money lost in every melt.

Why the usual fix fails. Shifting melts by hand into cheaper hours conflicts with the moulding line, which needs metal at a set time. Holding metal hot longer wastes what the shift saves.

How a fixed-price remote diagnosis checks it. From furnace logs, the pouring plan and the power tariff I check how a melting schedule against price and peak limit would have run, with metal ready when the line needs it. The fixed-price remote diagnosis shows whether this schedule pays in your foundry and what it would need from planning.

Background: Economic predictive control

Cupola furnace: coke rate and metal temperature set by experience

What happens. The cupola crew sets coke and blast by experience and changes them when the tapped iron looks too cold or the analysis is off. The furnace state inside is not measured.

What it costs. Coke is a main cost of a cupola, and too much of it also adds to emissions. Too little gives cold iron and poor quality, so crews err on the high side.

Why the usual fix fails. Experience is carried by a few people, and the furnace reacts slowly, so changes by feel arrive late. Tapped-iron temperature measures the past, not what is now inside.

How a fixed-price remote diagnosis checks it. From charge records, blast data, tap temperatures and lab analyses I estimate the furnace state and check whether predictive control on that estimate could have saved coke at the same iron temperature. The fixed-price remote diagnosis tells you what such control would need on your cupola.

Background: Predictive control of process plants

Metal cools in the ladle while waiting to be poured

What happens. Metal is tapped, treated and then waits for the mould line. The longer it waits, the colder it gets. The pouring temperature is measured at the start, if at all, and then the ladle moves on.

What it costs. Cold shuts and misruns when metal is poured too cold, and extra superheating, with the energy that costs, to compensate. Rejects are found only after fettling or machining.

Why the usual fix fails. Tapping hotter is the standard answer and costs furnace energy and refractory life. A thermocouple dipped before pouring shows one value after the loss has happened.

How a fixed-price remote diagnosis checks it. From tap temperatures, waiting times, ladle data and pouring temperatures I check whether the temperature of the metal in the ladle can be computed during the wait. The fixed-price remote diagnosis answers how accurate that value can be and whether it could set tapping temperature case by case.

Background: Computed measurements

Moulding sand properties drift

What happens. Moulding sand is reused and changes with every cycle: moisture, binder, fines and temperature move. The mixer is set to a recipe, and the lab tests samples some time after the sand has been used.

What it costs. Casting defects that come from the sand: scabs, gas holes, poor surface. Finding the cause takes days, and the sand has changed again by then.

Why the usual fix fails. Raising water or binder additions as a safety margin wastes material and shifts the problem. The lab test arrives too late to steer the mixer.

How a fixed-price remote diagnosis checks it. From mixer logs, sand temperatures, moisture sensors, addition records and lab values I estimate the sand state between tests and check whether dosing of water and binder could follow it. The fixed-price remote diagnosis shows how much of the sand drift can be seen from data you already record.

Background: Computed measurements

Heat treatment of castings runs on fixed long cycles

What happens. Castings go into the furnace with a standard cycle that is long enough for the heaviest load. The thermocouple measures the furnace, not the casting, and the cycle runs to its end whatever the load.

What it costs. Gas for hours that the parts do not need, and furnace capacity for the next load. Cycles that are too short would risk the mechanical properties, so nobody shortens them.

Why the usual fix fails. A measurement thermocouple on selected parts shows one part. A cycle shortened by hand holds until a heavy or dense batch arrives.

How a fixed-price remote diagnosis checks it. From furnace logs, load descriptions and any measured part temperatures I compute the part temperature over the cycle and check how much of the hold time was really needed. The fixed-price remote diagnosis shows whether a computed part temperature can end the cycle and what that requires.

Background: Computed measurements

Castings cool too long before shakeout

What happens. After pouring, castings stay in the mould for a fixed time before shakeout. The time is set so that even the biggest and hottest casting is cool enough. Most are cool earlier.

What it costs. Throughput: moulds are blocked, the line waits and the cooling area is full. Shakeout too early, on the other hand, causes distortion and cracks.

Why the usual fix fails. A single cooling time for all patterns is easy to run. A temperature measurement of each casting in the mould is not possible, so the time stays conservative.

How a fixed-price remote diagnosis checks it. From pouring temperatures, pattern data, mould times and any measurement on the sand I estimate how fast each casting cools and what time each pattern needs. The fixed-price remote diagnosis tells you how much cooling time could be saved on which patterns and what data you would have to log.

Background: Computed measurements

Scrap is found only after machining

What happens. Porosity, inclusions and shrinkage often show only when the machine cuts the casting open. By then the foundry has paid for melting, moulding, cleaning and a good part of the machining.

What it costs. Cost of waste along the chain, and complaints from the customer’s machine shop. Finding the cause takes time because the melt and pouring conditions have been forgotten.

Why the usual fix fails. More inspection of finished castings finds the scrap earlier in the process but still after the fact. Tightening process limits across the board slows everything.

How a fixed-price remote diagnosis checks it. From melt, pouring and moulding data matched to scrap records I check whether the likelihood of a defect can be estimated for each casting before it leaves the foundry. The fixed-price remote diagnosis shows whether your records are enough for that and which measurement is missing.

Background: Computed measurements

Alloy corrections are made from samples

What happens. The melter takes a sample, waits for the spectrometer, adds alloying elements and takes another sample. Between the samples the composition is a guess.

What it costs. Time with a furnace waiting, and material, because corrections overshoot or are added in small steps. Out-of-spec metal that is poured costs more than the correction.

Why the usual fix fails. Sampling more often takes the melters’ time and still leaves gaps. Rules of thumb for additions depend on the charge, whose scrap mix varies.

How a fixed-price remote diagnosis checks it. From charge records, additions, melt losses and spectrometer results I check whether the composition can be computed between samples and how well the next addition could be sized. The fixed-price remote diagnosis shows how many samples such a calculation could save and what accuracy to expect.

Background: Computed measurements

Compressed air, cooling and furnaces are not measured

What happens. In many foundries only the main electricity meter and the gas bill are available. Compressed air, cooling water, furnaces and auxiliary equipment are not metered separately, so nobody knows who uses what.

What it costs. Hidden losses: leaks, idle machines left running, cooling that runs all night. Without numbers there is no list of measures to rank.

Why the usual fix fails. A full set of submeters is an investment and a project in itself. Walk-through audits give a snapshot of one day.

How a fixed-price remote diagnosis checks it. From the main meters, production data and the few submeters you have I build an energy balance model that splits consumption across consumers and shows where the largest losses must be. The fixed-price remote diagnosis shows which consumers deserve a meter and where control could save energy.

Background: Economic predictive control

Skilled melters retire and the know-how sits in two heads

What happens. The experienced melters and furnace operators know how to judge a melt by the colour of the slag, the sound of the furnace and the way the metal flows. Two or three of them hold this knowledge, and they are close to retirement.

What it costs. The risk is a drop in quality or a longer ramp-up when they leave. Trainees need years to build the same feel, and mistakes show up as scrap.

Why the usual fix fails. Writing the know-how down as instructions loses what is not put into words. Hiring an outside expert for each problem leaves nothing behind in the plant.

How a fixed-price remote diagnosis checks it. I take melt logs, analyses, additions and the experienced melters’ notes and rebuild their rules as a model that explains what the data shows. The fixed-price remote diagnosis shows how much of the know-how can be captured in a documented model that assists new operators.

Bigger castings with the same process window

What happens. Customers want larger and heavier castings, but the foundry’s furnaces, ladles and moulding line stay as they are. The process window for pouring, feeding and cooling gets narrower with the size.

What it costs. Defects such as shrinkage and misruns, and trial pours on expensive patterns. Every new size starts with scrap.

Why the usual fix fails. The settings from smaller parts are carried over and corrected after each failed pour. Simulation of solidification alone does not capture the real furnace, ladle and mould line.

How a fixed-price remote diagnosis checks it. From past pours, process data and defect records of the existing sizes I check how the process behaves as size grows, and what a model for the new size must include. The fixed-price remote diagnosis shows whether the plant can reach the new size within its window and what to change first.

Background: Control that adapts to drift and wear

Core curing time is fixed

What happens. Cores are hardened in a box or an oven for a fixed time. The time is chosen so that the least favourable core is hard, and every core gets it.

What it costs. Throughput of the core shop, gas or amine use, and cores that are still soft in the centre if the setting is wrong.

Why the usual fix fails. Shortening by trial is a risk: a core that breaks in the mould shows up as scrap far later. A hardness test on a core gives a sample, not a value for each.

How a fixed-price remote diagnosis checks it. From curing records, temperatures, gas flows and core dimensions I estimate how far each core has hardened and check how much curing time could be saved. The fixed-price remote diagnosis shows whether a computed cure state can end the cycle.

Background: Computed measurements

Holding furnaces are kept hot all weekend

What happens. Holding furnaces stay at temperature over weekends and holidays because nobody wants to find them cold on Monday morning. The metal in them may be gone, but the burners or heating continue.

What it costs. Power or gas for hours with no production, and refractory ageing from thermal cycling when a furnace is restarted too fast.

Why the usual fix fails. A fixed weekend programme is either too safe or too risky, depending on when the next pour is due. Switching off and on by hand relies on someone remembering.

How a fixed-price remote diagnosis checks it. From furnace logs, heat-loss curves and the production plan I compute how far the furnace can cool and when it must restart to be ready. The fixed-price remote diagnosis shows how much weekend energy a plan-led holding strategy could save and what it takes to run it safely.

Background: Economic predictive control

Exhaust and filter fans run at full speed

What happens. The fans of the dust extraction run at full speed all shift, whether the furnaces are charging, tapping, or idle. Dampers are set once and rarely touched.

What it costs. Fan power is a steady and large part of the electricity bill. Too little extraction at the wrong moment, on the other hand, is a problem for air quality and for permits.

Why the usual fix fails. Throttling with dampers wastes energy in the damper. A fixed lower speed risks the worst moment of the cycle.

How a fixed-price remote diagnosis checks it. From fan power, pressure and flow readings and the melting cycle I check how demand-led control of the fans would follow the extraction need of each source. The fixed-price remote diagnosis shows how much of the fan energy is demand-led in your plant and what instrumentation the control needs.

Background: Economic predictive control

Pattern changes and melts are not aligned

What happens. Moulding changes the pattern on its own schedule, while the melt shop produces metal in batches on another. The two are matched by the planner and by phone calls.

What it costs. Idle furnaces waiting for a pattern change, or a moulding line waiting for metal, and extra holding of liquid metal. Both are costly in energy and capacity.

Why the usual fix fails. Better scheduling of each shop separately does not solve the interface. A planner with a spreadsheet cannot test many sequences quickly.

How a fixed-price remote diagnosis checks it. From the production plan, melt times, pattern change times and furnace logs I check how an integrated schedule of melting and moulding would have run. The fixed-price remote diagnosis tells you how much idle time and holding energy a joint schedule could remove, and what planning data it needs.

Background: Optimal paths and setpoint plans

Thermocouples in harsh areas keep failing

What happens. Thermocouples in furnaces, in ladles and in hot, dusty areas fail often. When one fails, the control system has no value and the operator runs blind or falls back on a fixed setting.

What it costs. Downtime for replacement, uncontrolled temperatures, and a bad reading that nobody notices can be worse than none.

Why the usual fix fails. A spare thermocouple next to every one doubles cost and maintenance. Detecting a drifting reading by eye relies on the operator knowing what it should be.

How a fixed-price remote diagnosis checks it. From temperature and energy logs I check whether a failed or drifting reading can be detected by comparing it with a model, and whether the model can stand in until it is replaced. The fixed-price remote diagnosis shows which of your measurements can be covered this way.

Background: Computing unmeasured states for control

Moving from gas to electric melting

What happens. A foundry replaces gas-fired furnaces by electric ones or adds induction furnaces. The new furnace heats, cools and reacts differently, and the operators know the old one.

What it costs. Teething problems: melt times that do not match the plan, higher peaks on the grid connection, and unfamiliar refractory wear.

Why the usual fix fails. The control system of the new furnace is delivered with default settings, tuned for a standard case, not for the foundry’s charge, batch plan and tariff.

How a fixed-price remote diagnosis checks it. From the data of the existing furnaces, the new furnace’s specification and the foundry’s plan I work out how the new furnace should be run and what control it needs. The fixed-price remote diagnosis shows what the concept for the new furnace would look like before it is installed.

Background: Predictive control of process plants

Public plans for all economic energy savings need credible figures

What happens. Larger industrial energy users must draw up and publish plans for the economically sensible savings measures found in their energy audit, and an auditor has to confirm them. Deadlines are approaching, and an amendment could shorten them.

What it costs. A public, checked plan with figures that cannot be defended is a risk. Producing it from scratch ties up people who are already short of time.

Why the usual fix fails. Rough estimates from a consultant’s walk-through are quick but are hard to defend. Counting savings from submetering needs data the plant does not have.

How a fixed-price remote diagnosis checks it. From process data, production plans and plant documentation I compute savings for the candidate measures from the energy audit and check how well the figures stand. The fixed-price remote diagnosis shows which savings can be backed with your data and which need more measurement.

The carbon price on gas rises and stays uncertain

What happens. Gas for process heat carries a carbon price that is meant to rise, and the start of the European emissions trading for heating fuels has been postponed by a year (Clean Energy Wire). Foundries with gas-fired furnaces and heat treatment cannot plan with it.

What it costs. A rising and uncertain price on gas decides whether electrification or heat recovery pays. Investing too early or too late costs real money.

Why the usual fix fails. Rules of thumb for payback ignore the real operating pattern of furnaces and the way energy prices move. A study for one price scenario is out of date when the price path changes.

How a fixed-price remote diagnosis checks it. From energy data and operating schedules I build a model of your gas and power use and compute the case for electrification and heat recovery across several price paths. The fixed-price remote diagnosis shows which investments hold up under uncertain prices.

Background: Economic predictive control

Subsidies for measurement, control and process optimisation go unused

What happens. Federal funding exists for investments in measurement and control technology, energy-management software and the optimisation of plants and processes in companies (federal funding database). Many foundries do not know it covers projects of this kind.

What it costs. Projects that would pay are postponed because of the cost, and the funding is not claimed.

Why the usual fix fails. A project written as a general consultancy does not fit the funding rules. Preparing an application takes time that the plant manager does not have.

How a fixed-price remote diagnosis checks it. I shape the content of the diagnosis so that the measurements and models it calls for can be described in the terms the funding uses. The fixed-price remote diagnosis gives you a defined scope and a result in writing. Whether a project qualifies is for you to check against the current programme conditions.