Plant builders: start-up takes weeks of trial and error

Start-up takes weeks of trial and error

What happens. The plant is built and the control is loaded, but the real behaviour shows only at the first full run. Loops oscillate, sequences stop at conditions nobody thought of, and the commissioning team tunes one loop after the other on site.

What it costs. A late hand-over, penalties in the contract and a team that stays on site far longer than planned. Faults found at the first full run are expensive to fix.

Why the usual fix fails. More people on site speed up the tuning only a little, as the plant can only be tested one step at a time. Tuning rules of thumb fail on loops with long delays and interaction.

How a fixed-price remote diagnosis checks it. From the process description, the design data and the control documentation I check which loops carry the risk and whether a dynamic model can give start values for the tuning before the first run. The fixed-price remote diagnosis tells you what such a model would cost in effort and how much site time it could save.

Background: Predictive control of process plants

Experienced commissioning engineers are scarce and travel for weeks

What happens. The few engineers who can bring a plant to stable operation are booked far ahead and travel to the customer’s site for weeks, often abroad.

What it costs. Travel cost, delays when nobody is free, and start-ups led by less experienced people with more trial and error.

Why the usual fix fails. Hiring such engineers takes years. Written procedures cover the standard case, but not the judgement needed when the plant does something unexpected.

How a fixed-price remote diagnosis checks it. From the plant data sent in during start-up (trends, alarms, tuning settings) I look at what is going wrong, remotely, and check which loops could be pre-tuned on a model. The fixed-price remote diagnosis tells you how much of the expert work could be done from a distance and where a person on site is still needed.

The control concept is fixed late in engineering

What happens. In the engineering phase, the process is designed and equipment is ordered, but the philosophy for how it will be controlled is left for later. When it is written, the plant layout already limits what is possible.

What it costs. Changes late in the project, controls that fight the design, and extra work in commissioning.

Why the usual fix fails. Standard control concepts are copied from earlier projects and do not fit the new process. Specialists for control concepts are hard to find in the project team.

How a fixed-price remote diagnosis checks it. From the process flow diagram, the equipment data and the operating cases I check which parts of the plant need more than standard loops and what architecture would suit them. The fixed-price remote diagnosis tells you where the control concept needs attention early.

Customers expect process optimisation at hand-over

What happens. More customers want the plant not only to run but to run well, with digital process optimisation set up at their site during commissioning. The plant builder needs people for control strategy and data analysis.

What it costs. Delays when the right people are missing, and a hand-over that satisfies the order but not the customer’s expectation.

Why the usual fix fails. Software suppliers deliver tools but no one who knows the process. The plant builder’s own team is busy with the start-up.

How a fixed-price remote diagnosis checks it. From a plant’s design and early operating data I check which optimisation tasks are realistic and what they need in measurements and models. The fixed-price remote diagnosis tells you what you could offer your customers at hand-over.

Background: Economic predictive control

Simulation before start-up is wanted but rarely done

What happens. Virtual commissioning and simulation models for machines and plants are in demand, but building them is a project of its own. In many companies, they stay on the wish list.

What it costs. Faults that could have been found on a model show up on site, where each hour is costly.

Why the usual fix fails. Detailed simulation models take too much effort and often come too late. A model that is too rough shows nothing useful.

How a fixed-price remote diagnosis checks it. From the documentation of a typical plant I check what level of model suffices to test the control logic and the tuning before start-up. The fixed-price remote diagnosis tells you what a simulation for your type of plant would need and what it would find.