Chemical parks: utilities run plant by plant, not as one site

Steam header pressure swings when large consumers switch

What happens. A tenant starts a batch, a column trips or a turbine goes off line, and the steam demand on the header jumps. Boilers, letdown stations and turbines each have their own pressure controller, and they answer one after another. The pressure overshoots, safety valves lift or steam is vented, and sometimes a sensitive consumer trips.

What it costs. Steam vented to atmosphere, extra fuel in boilers kept on hot standby, and trips that cost tenants a production run. Disputes over who caused what.

Why the usual fix fails. More spinning reserve in the boilers and wider pressure bands make the network calmer but cost fuel every hour. Retuning single controllers just moves the problem.

How a fixed-price remote diagnosis checks it. From your header pressures, boiler loads, letdown and turbine flows and the consumption records of large users, I look at how disturbances travel through the network and which units fight each other. The fixed-price remote diagnosis tells you whether coordinated predictive control of boilers, turbines and letdowns could hold the pressure with less reserve, and what it would take.

Background: Predictive control of process plants

Cooling water pumps and tower fans run flat out whatever the load

What happens. The site cooling water system was sized for all plants running at full load on a hot summer day. Pumps run at fixed speed, tower fans are on or off, and the supply temperature is kept low all year. On a cool night with half the plants at part load, the system still runs as if it were July.

What it costs. Electrical power for pumping and fans that nobody needs, and pumps running far from their best point, which costs wear too.

Why the usual fix fails. Each tenant wants guaranteed supply pressure and temperature, so the operator keeps margin. Fitting variable speed drives helps only if someone decides how far the setpoints can safely move.

How a fixed-price remote diagnosis checks it. From your pump and fan data, supply and return temperatures, pressures, flows to the main consumers and weather data I check how much cooling is really needed and where the limits are. The fixed-price remote diagnosis tells you whether load-led control of pumps and towers could save power while keeping every tenant’s supply, and what it would need.

Each plant is optimised on its own; the site as a whole is not

What happens. Each tenant optimises its own plant, the utility operator runs boilers, turbines and the grid connection, and another team buys energy. Steam, power, cooling water, nitrogen and compressed air flow between them at fixed prices or by contract. No one looks at which combination is cheapest for the site at a given hour.

What it costs. Steam reduced through valves while a turbine could have made power, gas bought while waste heat is cooled away, plants running at peak price that could have shifted.

Why the usual fix fails. Energy audits find savings, but each needs several parties to change how they operate. Without a shared model of the site, the decision is never made.

How a fixed-price remote diagnosis checks it. From your utility data, tenant consumption profiles, contracts and price data I map the flows and where a site-wide schedule could do better. I worked on a similar coordination question in predictive control and state estimation for the superfluid-helium cooling of the Large Hadron Collider at CERN (the European particle-physics laboratory). The fixed-price remote diagnosis tells you what site-wide economic optimisation could be worth and what it would take.

Background: Economic predictive control

Grid fees rise for sites with their own line or substation

What happens. The federal grid agency has ended the individual grid fees that sites paid for lines and substations used only by them; the rule stops applying from 1 January 2026 (Rödl & Partner). Sites with their own connection now face the general fees, where the peak load counts.

What it costs. Higher grid costs, and a peak that matters more than before: a single quarter-hour with all large drives and electrolysers on can set the bill for a whole year.

Why the usual fix fails. Peak alarms warn the control room, which then sheds load by hand, usually too late or at the cost of production. Tenants do not see the site peak at all.

How a fixed-price remote diagnosis checks it. From your load profile, the large consumers and generators on site, storage and the tenants’ flexible processes I check where peaks come from and which loads could move. The fixed-price remote diagnosis tells you whether scheduling the flexible loads could cut the peak and the fees, and what it would need.

Steam is raised from gas while power is cheap or even negatively priced

What happens. On sunny and windy days the power price falls, sometimes below zero. The site’s steam still comes from gas boilers or combined heat and power, which may even produce power into the grid at that moment. An electric boiler or heat pump could use the cheap power, but there is none, or it is not run to the price.

What it costs. Gas burned and carbon paid for while electricity is nearly free, and power sold at a loss when it is in surplus.

Why the usual fix fails. An electric boiler on its own does not help if it is switched by hand on a few obvious days. Without storage and a plan for the whole steam system, most cheap hours pass unused.

How a fixed-price remote diagnosis checks it. From your steam demand, boiler and combined heat and power data, power prices or contracts and the options for storage, I check how many hours an electric boiler or heat pump would really run and what it would save. The fixed-price remote diagnosis tells you whether price-led dispatch would pay and which equipment it needs.

Background: Economic predictive control

An electric boiler joins the steam system; nobody plans which source runs when

What happens. Electrode boilers are being installed in chemical parks so steam can be made from power when renewables are plentiful. They can change load very fast, much faster than the steam network can take. Now the control room must decide, hour by hour, between the electric boiler, combined heat and power and gas boilers, while keeping header pressure.

What it costs. An expensive asset that runs less than it could, or runs at the wrong times. Pressure swings when it ramps; gas boilers kept warm as backup.

Why the usual fix fails. Fixed priority lists and ramp limits ignore price, demand forecast and the state of the other units. Planning by hand each day cannot keep up with prices that change every quarter-hour.

How a fixed-price remote diagnosis checks it. From the data of all steam sources, header pressures, steam demand and price history I check how the sources should share the load and what limits the network sets on ramping. The fixed-price remote diagnosis tells you whether economic predictive control of the whole steam system would pay, and what it would take.

Tenants close plants, and shared utilities run at part load for fewer users

What happens. Large plants in chemical parks close or are cut back: crackers, chlor-alkali units, polymer plants. The park’s boilers, power plant, cooling water, nitrogen and wastewater treatment were sized for all of them. Now they run at part load, and the fixed costs are shared by fewer tenants.

What it costs. Worse efficiency at part load, higher prices for the remaining tenants, and the risk that they leave too. Some plants also lose a supply they took from the closed neighbour.

Why the usual fix fails. Running the old units at a lower load keeps everything working but at poor efficiency. Taking units off line by feel risks supply security.

How a fixed-price remote diagnosis checks it. From your utility data, the remaining tenants’ demand and the planned closures I check which units should run, at what load, and what supply security requires. The fixed-price remote diagnosis tells you what a re-optimised utility system for the smaller site could look like and what it would take.

New tenants bring first-of-a-kind plants with unknown loads

What happens. To fill empty plots, parks invite start-ups and new businesses, for example in recycling or new materials. Their plants are often the first of their kind. How much steam, power, cooling water and nitrogen they will draw, and how much it will swing during start-up, nobody can say from experience.

What it costs. Utility capacity reserved for loads that may never come, or supply problems when a new plant starts up with peaks nobody planned for. The new tenant’s start-up problems can disturb its neighbours.

Why the usual fix fails. Design values from the new tenant are optimistic or very conservative. Planning with large margins blocks capacity and money.

How a fixed-price remote diagnosis checks it. From the new plant’s design data and process diagrams, and from your present utility data, I estimate likely demand profiles during start-up and normal operation, and check how the site systems will cope. The fixed-price remote diagnosis tells you where the risks are and what control and planning would keep the park stable.