The KIKI project aims to enrich current inspection procedures for the maintenance of sewers with AI methods so that automated damage detection in image data becomes possible.
1
2
Furthermore, it is investigated to what extent a prognosis model based on historical data can predict the future aging process and whether an efficient maintenance strategy can be derived on the basis of this information.
3
For this purpose, the data is collected on a digital twin, which represents the condition of the sewer system in three dimensions and can be used for planning and control processes as well as for specific rehabilitation measures by means of mixed reality.
In a nutshell
What?
The KIKI project aims to enrich current inspection procedures for the maintenance of sewers with AI methods so that automated damage detection in image data becomes possible. This saves time and costs. In addition, it will be investigated to what extent a prognosis model based on historical data can predict the future aging process. An efficient maintenance strategy can then be derived on this basis.
How?
To support and accelerate the process, an automated evaluation of the TV image data will be combined with an AI-based damage detection. In order to predict the expected damages more precisely in the future, the database of a sewer database will be plausibilized and completed with an AI-based degradation model. A virtual model of the sewer network will then be created using VR/AR visualization (digital twin) based on this acquired data. This will enable effective access to condition and forecast data for experts to support rehabilitation decisions, maintenance proposals and maintenance planning.
Initial situation
The sewer network is a central component of the residential and industrial infrastructure in Germany. The sewer network is subject to a natural aging process which, without continuous care and maintenance, can lead to functional failure and even – in the case of leaks – to contamination of the groundwater. The speed of this aging process is subject to a variety of factors, such as the building fabric, the surrounding soil, but also the stress of traffic passing over the sewer stops. The assessment of this aging process and the decision to counteract it by sewer renewal, sewer rehabilitation and sewer repair is the responsibility of the operator of a sewer network and is based on an extensive stock of master data (location, material, year of construction, stationing, hydraulic load, manufacturing costs) and condition data, which are described by means of damage patterns and damage abbreviations and are managed in sewer databases.
Existing statistical forecasting models allow to predict the investment needs in a sewer network, but they require a very consistent master files and condition data stock and an extensive sample. Without a consistent and, as far as possible, complete master data stock, a meaningful prediction of the extent to which damage is imminent is not possible by means of qualified sampling. In practice, it has been shown that there are considerable data inconsistencies and data gaps in almost every sewer database, which must first be closed or checked for plausibility before a forecast model can be used.
sewer network as of 2018:
594,321 kilometers
cost approx.
2.500 € per kilometer
The August-Wilhelm Scheer Institute develops three basic components:
Automated detection and classification of damage in image data using AI methods.
Establishment of a data platform for the storage of condition data
Creation of a digital twin of the sewer network and provision in an AR application
Press releases
The August-Wilhelm Scheer Institute and the Entsorgungsverband Saar say they are working on a new solution for faster and more economical inspections of wastewater plants.
Recyclingmagazin.de, 31. August 2020
FUNDING NOTICE
The KIKI project is funded by the German Federal Ministry of Education and Research.
Funding code: 02WDG1594
Duration: 01.05.2021 bis 30.04.2023
KIKI is one of the Digital GreenTech projects. The funding guideline “Digital GreenTech – Environmental Technology Meets Digitalization” was published as part of the action plan “Natürlich.Digital.Nachhaltig”.