Skip to main content

Weather information utilising prediction models for the green density of pulpwood

Measurement of the pulpwood at pulp and paper mills is mainly carried out by using weight sampling, which is a robust and reliable measurement technology for large quantities of wood. The measured weight of timber is converted into volume using a conversion factor, defined as ‘green density’. The conversion factor is determined based on sample measurements. Sampling-based weighing has until now been mill-specific, and no universally accepted model has existed. 

The model is as good as the data it is built upon

Natural Resources Institute Finland (Luke) undertook a two-year R&D-project (2019–2021) in cooperation with Metsäteho Oy, UPM Oyj, Metsä Group and Stora Enso Oyj. The aim of the project was to develop weight-based measurement of pulpwood, so that the measurement methodology and the related determination of green density values are carried out rationally and in a way that is uniform. An addition aim was to utilise weather information in the estimation of green density. 

To build prediction models, data from almost 60,000 weight sample batch observations measured between 2013 and 2020 were collected from 17 pulpwood mill sites situated in different parts of Finland. The weight sample batch data was connected to weather information from the whole of Finland, provided by the Finnish Meteorological Institute.  

Utilising the data, the prediction models were built to estimate the green density factor values based on weather observations at the storage location and during the storage time. Relevant factors affecting green density include the time of the year, duration of storage, and based on weather observations, the mean temperature and precipitation during storage. Prediction models are available for five pulpwood assortments: pine, spruce, decayed spruce, birch and aspen. The models can be calibrated. 

Photo: Erkki Oksanen

Prediction models enable a new operating method

In the new operating method, the greenwood density of pulpwood arriving at the mill is estimated batch-specifically with the prediction models. The prediction models’ accuracy is continuously monitored and calibrated based on sample measurements. The aim of the continuous calibration is to correct the estimates to match observed values (e.g. year to year variation). 

 The sampling method was developed by Luke, and the selection and control of sampling batches is carried out across mill and company boundaries. This is a significant change to operating practices, as currently, sampling is performed at each mill separately. 

Changes to orders and practices

One of Luke’s statutory duties is to issue regulations on the conversion factors and functions to be used when measuring timber. This research led to the green density prediction models being formalised by Luke's order (1/2021), and made available for wood measurement from the beginning of 2022. 

In practice, the utilisation of the models requires the construction of a calculation service. The building of the Pulpwood Online (PWO) calculation service was launched in cooperation between TietoEVRY, MetsäGroup, StoraEnso, UPM, Metsäteho and Luke, in autumn 2021. PWO is scheduled to be operational and tested by the end of 2022. The users and beneficiaries of the new operating method will include large companies participating in its development, as well as SME-level operators and private forest owners. 

“Weather-based forecasting models for pulpwood exemplify how excellent results can be achieved through collaboration between research and practice organisations. The models’ versatility, usability and accuracy exceeded my expectations. They enable us to determine the volume of fresh density per batch, based on actual conditions, using weight sampling, which is Finland's most widely used timber measurement method.” Quality and Change Manager Timo Saarentaus, Metsä Group