Digitalisation

Digital transformation is currently taking place in all areas of public drinking water supply. Our expertise at TZW is in solving complex questions and problems of water supply companies. This is done either through direct technical support of companies or within the framework of practice-oriented research projects. In the process, all areas of drinking water supply - starting with resource protection, through water catchment and treatment, to distribution - are taken into account. Today, digitalisation plays a decisive role in all of these process steps. With our digital know-how, we can provide competent support here. This includes

• Process analysis and optimisation,
• the processing and evaluation of high-dimensional data using both deterministic and machine learning models
• technical-scientific expertise regarding the construction and maintenance of data architectures and databases
• as well as the development of solutions (web applications, software) for employees and customers.

Water demand forecast

The estimation of future water demand involves the inclusion of a variety of very dynamic components such as weather, urban and land developments in the supplied regions. TZW offers a web service for forecasting short-term water demand based on a transparent machine learning model. The water supply companies can access it directly via web.

Evaluation of water quality and quantity data

In all areas of water supply today, extensive data sets are created on a daily basis. Examples are the monitoring of water quality with online sensors, the time-resolved recording of water discharge quantities and consumption as well as the analysis of water samples with complex methods such as non-target screening or spectroscopy. The increasingly large amounts of data pose new demands to a company. We support our customers in the analysis and evaluation of this data. This includes, for example

• event detection, i.e. the detection of unusual patterns, e.g. in time series of water quality data. Events can be caused by real changes in water quality or by faulty sensors.
• Extraction of regular, recurring patterns that describe one or more normal states, e.g. water consumption types. This creates the basis for a better understanding of the system and helps, for example, to analyse the water consumption behaviour of customers or customer groups.
• the detection of water losses in time series of water delivery or consumption measurements
• the evaluation of high-dimensional spectroscopic data, e.g. fluorescence spectra, with which organic compounds in water can be detected.

Creation of regression and classification models (machine learning, AI)

When analysing data, relevant information on a wide variety of topics can be obtained. For example, statements about the concentration of a certain substance can be derived from spectral data of a water sample. Another example is the estimation of the condition of a shut-off valve or the estimation of a flushing interval for a section of pipe.

In all the examples mentioned, models are used that provide a certain information (target value) on the basis of input data, e.g. the spectrum of a water sample or the indicators of a pipe section. In the case of a continuous target value, one speaks of regression, e.g. the concentration value. In the case of a categorical target value, it is referred to as a classification, e.g. the specification of a condition as good, medium or bad.

To what extent the creation of such models is possible, we can check for various applications from all areas of water supply within the framework of feasibility studies. This involves checking the data basis and data quality, creating models and evaluating them in terms of performance or accuracy. The models are based, among other things, on machine learning methods. TZW places particular emphasis on the interpretability and transparency of these models. This is achieved by using models that are as transparent as possible and modern analysis methods. This includes Explainable AI, which allows the analysis of complex models, e.g. neural networks.

Databases

We support our customers in managing their data in databases efficiently and securely. TZW has extensive knowledge of the architecture of database systems and develops individual solutions together with our customers within the framework of technical-scientific consulting services. The focus is on PostgreSQL, MariaDB and small, portable database systems such as SQLite.

Web and GIS applications

In an increasingly networked world, access to up-to-date and accurate information is crucial for water utilities. We offer web and GIS solutions to collect, analyse and present relevant data in user-friendly interfaces. With these tools, companies can effectively manage and optimise their drinking water distribution network. Our applications enable users to visualise and analyse important information such as pipeline networks, pumping stations, water quality data, customer data and much more in real time.