Wastewater plants: fixed blower setpoints, chemicals with margin

Blowers hold fixed oxygen setpoints

What happens. The blowers supply air to the biology and hold an oxygen setpoint that was chosen once and left. Load and temperature change over the day and the year, and the plant spends air when the biology does not need it.

What it costs. Aeration is the largest power consumer of a treatment plant (Umweltbundesamt). A setpoint that is too high wastes this power; one that is too low endangers the ammonium limit.

Why the usual fix fails. Hand-lowering the oxygen setpoint saves power until a load peak comes in, and then the limit is exceeded and the old value returns. Ammonium probes are fine but foul and need maintenance.

How a fixed-price remote diagnosis checks it. From your plant logs (oxygen, ammonium and nitrate probes, blower power, inflow, lab values) I check how much air the biology really needs through the day. The fixed-price remote diagnosis tells you whether aeration led by ammonium computed between lab samples, with predictive control on an inflow forecast, can lower the blower power, and what it would take.

Background: Soft sensor for effluent quality

Laboratory analyses take people and hours

What happens. Samples are drawn, carried to the lab and analysed by hand, and the result arrives hours later. The people who do it are few, and sick leave or a vacancy stops the series.

What it costs. Delay between sample and decision, and time of qualified staff for routine work. Control that waits for lab values runs blind in between.

Why the usual fix fails. Taking fewer samples saves time and loses sight. Buying more online analysers moves the work to maintenance and calibration.

How a fixed-price remote diagnosis checks it. From the lab record, the online probes and the plant data I check which lab values can be computed from signals that already exist. The fixed-price remote diagnosis tells you whether computed values can replace part of the sampling, and which analyses must stay in the lab.

Background: Soft sensor for effluent quality

Qualified staff are missing at small plants

What happens. A small plant has one or two people, perhaps shared with other duties, and the operator who knows the plant retires soon. The plant works until something unusual happens.

What it costs. Compliance risk, because a problem is noticed late, and strain on the staff who are on call. Training a successor takes years the plant does not have.

Why the usual fix fails. A full-time expert is not affordable for a small association. Contracting maintenance companies for every visit costs more than it saves and does not provide ongoing attention.

How a fixed-price remote diagnosis checks it. From the plant’s records and the signals it already sends I look at which tasks can be automated and which can be watched remotely. The fixed-price remote diagnosis tells you what automated control and remote monitoring would cover, and which duties must stay with a person.

The dry matter of the sewage sludge varies

What happens. The dewatered sludge arrives at the dryer or the incinerator with a dry matter content that changes from load to load. The dryer is set for the average, and the furnace feels the difference in heating value.

What it costs. Energy for the dryer, and unsteady combustion if the sludge is wetter than assumed. New process steps for phosphorus recovery add to the load on the sludge line.

Why the usual fix fails. A lab value for dry matter comes after the sludge has moved on. A fixed dryer setting for the wet case spends energy on the dry load.

How a fixed-price remote diagnosis checks it. From dryer and furnace logs, the dewatering data and lab dry-matter results I check whether the dry matter can be computed in time to lead the dryer. The fixed-price remote diagnosis tells you whether dryer and incinerator control on that value pays, and what it would take.

Background: Computing unmeasured states for control

Storm water and industrial loads arrive as peaks

What happens. In heavy rain the inflow multiplies within an hour, and industrial dischargers add loads of their own. The plant sees the peak when it arrives, and the biology has no time to prepare.

What it costs. Limit violations at the outlet, washed-out sludge and a long recovery afterwards. The cost shows in compliance and in the work of getting the biology back.

Why the usual fix fails. A bigger storage tank is a building project. Setting the plant to a cautious operation all the time costs power and chemicals on every dry day.

How a fixed-price remote diagnosis checks it. From inflow records, rain data and the outlet values I check how far ahead a peak can be foreseen and what the plant can do with that time. The fixed-price remote diagnosis tells you whether predictive control on an inflow forecast can protect the limits, and what forecast it needs.

Background: Economic predictive control

New permits and nitrogen and phosphorus targets tighten the limits

What happens. Discharge permits get stricter, and the revised European urban wastewater directive sets further targets for nitrogen and phosphorus (VKU). The plant has to meet them with the tanks it has.

What it costs. Plants run closer to the limit with less room for peaks. Running closer means more risk of a violation, and every violation is reported.

Why the usual fix fails. More margin by cautious operation costs power and chemicals. Extra tanks need money and years.

How a fixed-price remote diagnosis checks it. From the outlet values, the inflow and the probe signals I check how much room the plant actually has under the new limits. The fixed-price remote diagnosis tells you whether predictive control that manages the margin lets you meet them without building, and what it would take.

Background: Predictive control of process plants

Lab values come daily, the control is blind in between

What happens. Ammonium and nitrate are known from the lab once a day or from a probe that is sometimes out. In between, the control runs on assumptions.

What it costs. Limit violations that nobody saw coming, and chemicals and power spent to stay safe. A bad hour falls between two lab values.

Why the usual fix fails. More samples tie up the lab. Online probes foul and drift and must be cleaned and calibrated.

How a fixed-price remote diagnosis checks it. From the probes you have, the inflow and the lab series I check whether ammonium and nitrate can be computed continuously and checked against the lab. The fixed-price remote diagnosis tells you how well a computed value would bridge the time between samples.

Background: Soft sensor for effluent quality

Carbon source for denitrification is dosed with a margin

What happens. External carbon (an easily degraded source, usually bought in) is dosed to help denitrification. It is set by hand or by fixed ratio, with a margin so that the nitrate limit is not at risk.

What it costs. Chemical cost for carbon that the biology did not need, and more sludge. Too little carbon leaves nitrate at the outlet.

Why the usual fix fails. Lowering the dose by hand works until the load changes. A fixed ratio to flow ignores that nitrate and the carbon already in the water change on their own.

How a fixed-price remote diagnosis checks it. From dosing records, nitrate and inflow signals and lab values I check how much carbon the biology needs at which load. The fixed-price remote diagnosis tells you whether a dose based on the estimated demand can cut the chemical cost, and what probes it would rely on.

Background: Soft sensor for effluent quality

Precipitant for phosphorus is dosed generously

What happens. Iron or aluminium salt is dosed to remove phosphorus, often at a rate that stays safely above the limit. The dose follows the flow or a hand setting, and the actual phosphorus load comes in waves.

What it costs. Chemical cost, more sludge to dewater and dispose of, and wear on the dosing and the dewatering. An overdose does not show in the outlet, only in the bill.

Why the usual fix fails. Reducing the dose without a measure of the load risks the phosphorus limit. A phosphate probe at the outlet reports after the fact.

How a fixed-price remote diagnosis checks it. From dosing, flow, phosphate probes and lab values I check whether the load can be estimated at the inlet or in the tank and the dose led by it. The fixed-price remote diagnosis tells you what the margin is costing and whether load-based dosing is worth building.

Background: Computing unmeasured states for control

Nitrous oxide emissions are not measured

What happens. Biological treatment releases nitrous oxide, a strong greenhouse gas, and most plants do not measure it. What is reported in climate balances is a standard factor.

What it costs. Climate reporting that is built on a factor, and no knowledge of which operating states cause the emissions. If reporting becomes stricter, the plant has no evidence of its own.

Why the usual fix fails. A gas analyser on every tank is expensive and needs maintenance. A standard factor says nothing about what the plant could do to reduce the emission.

How a fixed-price remote diagnosis checks it. From the plant’s signals (oxygen, nitrate, nitrite where measured, load) and short measurement campaigns, if you have them, I check whether the emission can be estimated from operating states. The fixed-price remote diagnosis tells you how far an estimate carries and which operating conditions drive it.

Background: Soft sensor for effluent quality

Polymer for dewatering is dosed by feel

What happens. Polymer is added to the sludge before the centrifuge or the press. The operator sets the dose by looking at the cake and the filtrate, and the sludge changes from one hour to the next.

What it costs. Polymer cost, and more disposal cost when the cake is wetter than it could be. An overdose also clogs the machine.

Why the usual fix fails. A fixed dose per tonne of sludge ignores the solids content, which varies. A handheld test in the lab comes after the dose is already gone.

How a fixed-price remote diagnosis checks it. From dosing, sludge flow, solids probes and the dryness results of the cake I check whether the solids content can be estimated in time to set the dose. The fixed-price remote diagnosis tells you whether dosing on the estimated solids can lower polymer and disposal cost, and what it would take.

Background: Computing unmeasured states for control

Sludge is dried for the worst case

What happens. The sludge dryer runs at a setting that is enough for the wettest sludge. The drier loads leave the dryer drier than needed.

What it costs. Heat is spent on water that was not there, and the dryer capacity is used up for the next load. A product drier than required also changes how it burns.

Why the usual fix fails. Shortening the drying by hand risks sludge that is too wet for the furnace. A single moisture probe at the outlet shows one point after the dryer.

How a fixed-price remote diagnosis checks it. From dryer logs (temperatures, throughput, outlet readings) and the lab values I estimate the dry matter along the dryer. The fixed-price remote diagnosis tells you whether a computed dry matter can lead the dryer and how much energy that would leave out.

Background: Computing unmeasured states for control

Phosphorus recovery is a new process step

What happens. Larger plants have to recover phosphorus from sludge or ash in the next years. The process is new for most plants, and its reference operation is still thin.

What it costs. A new process step to commission, with unknowns in its behaviour, and cost to be carried by the fees. Time pressure leaves little room for learning on the real plant.

Why the usual fix fails. Taking the supplier’s recommended settings is the start but not the end; the real sludge is different. Waiting for experience from other plants means a late start.

How a fixed-price remote diagnosis checks it. From the process data of the supplier, your sludge analyses and the plant layout I look at how the recovery step fits the sludge line. The fixed-price remote diagnosis tells you what the concept and the control of the new step should look like, and which tests can be done in simulation first.

Background: Replacing hand-tuned parameter tables by a model

Gas engines follow the gas level

What happens. The digester gas drives engines for power and heat, and the engines run when the gas store is full and stop when it is empty. They do not look at the power price or the heat demand.

What it costs. Lost value: power produced when it is cheap, not when it is dear, and heat produced when nobody needs it. The gas is also flared when the store is full at the wrong time.

Why the usual fix fails. A bigger gas store takes money and space. A schedule by clock time ignores that the gas production changes with the sludge feed.

How a fixed-price remote diagnosis checks it. From gas production, store levels, engine data, prices and heat demand I check how much a plan that looks ahead would earn. The fixed-price remote diagnosis tells you whether scheduling the engines and the gas store pays, and which forecasts it needs.

Background: Economic predictive control

Probes foul and drift

What happens. The probes in the tanks (oxygen, ammonium, nitrate, solids) foul and drift between cleaning rounds. The control reads values that are slowly more wrong.

What it costs. Wrong control: air and chemicals are dosed against a reading that is off, and nobody notices until the lab value disagrees.

Why the usual fix fails. More frequent cleaning costs staff time. Calibration against the lab happens once a day at best.

How a fixed-price remote diagnosis checks it. From the probe signals, the lab values and the rest of the plant data I check whether a model running alongside can flag when a probe drifts, and bridge when it fails. The fixed-price remote diagnosis tells you which probes it can watch and how much maintenance that could save.

Background: Control that adapts to drift and wear

Shock loads from industrial dischargers arrive unannounced

What happens. A dairy, a food factory or a chemical site discharges a load into the sewer, and the plant sees it only at the inlet. The biology reacts to something it was not warned about.

What it costs. Biology upsets and limit violations that take days to recover from, and a dispute about who caused what.

Why the usual fix fails. A contract with limits for the discharger is checked by sampling. Rigid control of the plant does not distinguish a shock from the normal swing.

How a fixed-price remote diagnosis checks it. From the inlet probes, the flow and the discharger’s known production times I check whether a shock load can be recognised early and what the plant can do in the minutes it has. The fixed-price remote diagnosis tells you whether estimating the load at the inlet is feasible and how the control would use it.

Background: Computing unmeasured states for control

Pumps follow level switches

What happens. Pumps in the pumping stations start and stop at fixed levels. They run when the level is reached, irrespective of the time or the load downstream.

What it costs. Power cost from pumping at inefficient points and at times when the power is dear, and starts and stops that wear the pumps. Peaks arrive at the plant as bursts.

Why the usual fix fails. Moving the switch levels shifts the problem. Variable-speed drives only help if someone steers them.

How a fixed-price remote diagnosis checks it. From pump logs, level and flow records and the power prices I check how the pumping stations could be run together. The fixed-price remote diagnosis tells you whether optimised pumping lowers the power and smooths the inflow to the plant, and what the pumps and levels allow.

Background: Economic predictive control

Odour complaints

What happens. Odour comes from the inlet works, the sludge store and the tanks, and it is worst under certain weather and certain loads. Complaints arrive from neighbours, and the plant gets the news late.

What it costs. Fines, a poor relationship with the neighbours and a risk to the permit. Dosing of chemicals against odour is expensive if it runs all the time.

Why the usual fix fails. Scrubbers and covers are costly and treat the symptom. Dosing by a calendar misses the days that matter.

How a fixed-price remote diagnosis checks it. From the plant data, complaint dates, weather and odour measurements, if any, I check whether the odour can be predicted from the operating state. The fixed-price remote diagnosis tells you whether prediction and dosing on demand are feasible, and what to record.

Background: Computing unmeasured states for control

New tanks, old control

What happens. A plant is extended with new tanks, and the old control is adjusted to fit. The structure of the control was made for the old layout, and the new part is bolted on.

What it costs. Instability and unclear responsibility between old and new parts, and a plant that runs worse than the sum of its parts. The extension does not reach its promised performance.

Why the usual fix fails. Adding rules for the new tanks to the old program works for a while and makes the program less understandable every year. Handing the whole control to the supplier of the extension fixes one part only.

How a fixed-price remote diagnosis checks it. From the old control, the plant data and the plans of the extension I build a model of the extended plant. The fixed-price remote diagnosis tells you what a control concept for the whole plant would look like and where the old logic should be replaced.

Background: Replacing hand-tuned parameter tables by a model

New staff learn on the plant

What happens. New operators learn the plant on the plant. They are shown the rounds, the tanks and the alarms, and they learn what happens when a setting changes by changing it.

What it costs. Mistakes that cost limit violations or biology upsets, and a long period in which the new person is not trusted alone at night.

Why the usual fix fails. Handbooks describe the plant as designed, not as it behaves. Learning from a colleague depends on that colleague having time.

How a fixed-price remote diagnosis checks it. From the plant data and the plan of the tanks I build a model that reacts to changes of settings as the plant does. The fixed-price remote diagnosis tells you whether a plant model used as a simulator is realistic enough for training, and what the model needs from your records.

Background: Replacing hand-tuned parameter tables by a model

Energy audits and energy neutrality become duties

What happens. The revised European directive asks larger treatment plants for regular energy audits and, over time, for a rising share of renewable energy (VKU). The plant must show where the energy goes and how to cut it.

What it costs. Staff time for the audit, and a weak basis if the plant does not know the split between aeration, pumping, heating and the rest. Consultant reports that arrive once do not change daily operation.

Why the usual fix fails. A one-off audit lists measures that nobody operates. Sub-metering everywhere is costly.

How a fixed-price remote diagnosis checks it. From your meter data, aeration and pumping logs and the digester and engine data I build an energy balance of the plant. The fixed-price remote diagnosis tells you where the largest saving lies, and whether an aeration model and a schedule for digester and engines are worth it.

Background: Economic predictive control

A fourth treatment stage for micropollutants

What happens. Larger plants have to remove micropollutants (traces of drugs and cosmetics) with ozone or activated carbon, and the makers of these products are to carry most of the cost (ZfK).

What it costs. Ozone and carbon cost power and money in proportion to the dose. The dose is set against the worst case, so the stage works at full strength all the time.

Why the usual fix fails. A dose proportional to flow ignores that the load of micropollutants changes. Direct measurement of micropollutants is not possible online.

How a fixed-price remote diagnosis checks it. From the stage’s logs, the flow, organic indicators and lab results I check whether the load can be computed and the dose led by it. The fixed-price remote diagnosis tells you what load-based dosing could save and which indicator signals it would rely on.

Background: Computing unmeasured states for control

Sludge disposal and phosphorus recovery are rushed

What happens. Larger plants must recover phosphorus from sewage sludge by a fixed date. Mono-incineration and recovery plants are needed in numbers, and their capacity is expected to fall behind what is required (zfk).

What it costs. Plants rushed into commissioning, a disposal route that is not secure and fees that rise. Whether a fund or a postponement comes is open.

Why the usual fix fails. Waiting for the final rules costs the years a plant needs. Commissioning with hurry and supplier settings means a long and costly run-in.

How a fixed-price remote diagnosis checks it. From the sludge data and the plans of the associations and builders I check how combustion, drying and recovery can be controlled from the first day. The fixed-price remote diagnosis tells you what simulation and a control concept can bring forward in time, and what the data must show.

Background: Replacing hand-tuned parameter tables by a model

The wastewater levy is billed on permitted loads

What happens. The levy is based mainly on the loads written in the permit, not on the loads the plant actually discharges (Abwasserabgabengesetz).

What it costs. Operators pay on permitted loads that they stay well below. The difference is a cost without a service behind it.

Why the usual fix fails. Declaring a lower load needs proof, and a lab series every few days is not enough proof. Waiting for the reform delays the saving.

How a fixed-price remote diagnosis checks it. From the operating data, lab series and the outlet probes I check whether the discharged loads can be computed well enough to support a lower declaration. The fixed-price remote diagnosis tells you how solid that proof would be, and what the authority would need to accept it.

Background: Computing unmeasured states for control