The Digital Sawmill is a project that has been going on for two years, 2017-2018. The project is funded by Vinnova and the participating companies Moelven Industrier, RemaSawco, Schneider Electric and RISE. The aim of the project was to, by digitizing the sawmill's production and process data, increase the process efficiency by 15%, increase the product value by 10% and reduce energy consumption by 10%.
Installations and practical work on site has been performed at Moelven Valåsen.
In the beginning of the project, a specification of needs and inventory of what already existed was made. An initial inventory of communication and data storage was carried out by Moelven's own IT department together with Schneider Electric. Inventory of sensors to measure on the wood was carried out by Moelven and RemaSawco.
To connect different sensors, to filter and give the possibility to choose which values to send and to a certain extent standardize the data formats all sensors and sensors are connected over an integration platform before data is sent to storage. In this way, with an integration platform, it is fairly easy to add and change sensors. Moelven's choice of integration platform fell on Apache NiFi.
Moelven Valåsen has several systems with local storage. With the help of integrations, the storage location is doubled against a cloud service, in our case Google Cloud. To reduce the storage requirement in Google's Cloud, data is filtered at integrations where only values of benefit for analysis are transferred.
Even before this project started, a large number of sensors were installed. They worked locally and solved local measurements such as measuring diameter and quality of logs as well as dimensions, quality and moisture ratio on planks. In addition, new sensors and sensors have been installed to measure both the processes and the products.
To achieve the set goals, some thirty analyzes were defined, all of which support one or more of the formulated objectives and which illustrate potential improvement opportunities. The analyzes are based on measurements of energy, power, measurements of logs, fingerprints on log compared to board, blocks and sawn / dried goods, as well as at times and time intervals in the process. Of the formulated analyzes, the project has developed several manual analyzes and results, i.e. analyzes that require a manual start to obtain a report. Continued work is required to implement automatic reports. This will be done in a follow-up project where several Notebooks with Pythons scripts are adapted for Google Dataflow for real-time analysis.
The project has reached a number of sub-goals. Some of them are confirmed in the project while others have been calculated. The reason why a number need to be calculated is that, in order to be confirmed, they also require some rebuilding of mechanics, which was not possible during this limited project time. One such example is to be able to confirm an estimated 10% energy saving by optimizing the use of the kiln dryers.
Meters and sensors for measuring energy and power were installed too late in the project for an acceptable base level to be established.
Since circumstances beyond the control of the project have had a significant impact on production, it is difficult to demonstrate that the project has affected efficiency in particular. The project partners are fully assured that the process efficiency can be increased by at least 15% when the analysis tool is in place, so the causes of stops and downtime can be found and remedied before problems arise.
One of the ways the product value has increased is to increase the volume yield by reducing the green dimensions. This is done through the possibility of checking both the green dimensions and the dimensions of the finished products. A calculation shows a potential increase in value of approximately SEK 5 million per year.
Another way to increase the product value is by means of correlating board quality to log quality. A method for identifying corresponding logs and boards was further developed and a large number of correlated log and board grades were collected. This information can contribute to a considerably improved timber grading and sorting, but within the time frame of the project there was no time to evaluate changed log sorting settings. The increase in product value based on this methodology can be estimated to at least 5%.
The project has clarified how digitization requires collaboration between different sensors, sometimes from different suppliers, in order to achieve the goals of the project. Such collaboration places demands on open system solutions and a well-considered overall system architecture.
Some important lessons learned from the project can be summarized in a number of points:
- Digitization is a complicated project that takes time and requires resources.
- It is essential to focus on the goals set.
- Careful planning is required of: plant inventory, installation in stages, demanding deployment with parallel systems for new automated connections.
- Digitization projects start with details, continue with details and even more detail, with clear goals of a result of shaping a unified unit.
- Massive skills challenge.
- Visualize more, with less monitoring screens.
- Take control of data flows and take ownership of your own data.
- Platforms that handle data need to be tested in a harsh environment before
- Different systems that measure time, energy or dimension need to be synchronized onsite, and with periodicity calibrated against "approved" similar.
- Traceability in saw lines can be achieved with redundant systems that measure length with the addition of timestamp.
Another project result is a new way of tracing products in a process line. The method is called Delta Time, and is patent pending. According to the method, products, in our case logs, blocks and planks, can be traced by following the time pattern for measurements on the product. This means that all types of measurement can be used to trace a product. For example, length in a process measurement can be traced to power in the next and thickness in the third measurement.
In the above-mentioned patent application there is also a new way of presenting measurement values on the screen, called AugLog. In the presentation, you see values from a large number of sensors placed on the product being displayed. The measured values have been generated at different times and in different locations, but they are displayed together with the object to which they belong, which gives an overall picture of the measurements made on the product.
One very important project result is that we have formulated a method that describes how a digitization of an industry can go. Our experience from this project shows that it takes time to implement a digitization, and 2 years is a reasonable time frame to get a good bit on the way.