BUSINESS INTELLIGENCE SYSTEM FOR PLANTS

The customizable system integrates all the data about the production process into an easy-to-use web application, making cost analysis and planning of different scenarios much easier.

Customer

The client provides complex ERP systems for large enterprises, as well as audit and consulting services for manufacturing companies, helping them to optimize their work and financial statements. Detailed information about the client cannot be disclosed under the provisions of the NDA.

PROBLEM AND OBJECTIVES

Analysts of the client used to calculate all the costs and expenditures of different factories in cumbersome Excel spreadsheets. The more data were in the tables, the slower the program worked. Moreover, if it was necessary to add columns to the table (e.g. with new components), users had to change formulas within cells manually. That slowed down the analytical process and caused some human errors. In order to streamline the work and increase efficiency, the client came up with an idea to integrate an additional module in the existing ERP system, with detailed production analytics and forecasts. The module is supposed to be easily customized to any factory and production process, so users just add required data about machines, materials, components, and staff, while all the calculations are predetermined and proceed in the backend.

Technologien

SOLUTION

Our team developed a new part for an existing ERP platform from scratch, which covers the production structure, interrelationships between its components, cost calculation, and simulation. The first part of the solution represents a scheme with a plant structure. Users can build a custom model of their factory, adding level by level – workshops, equipment, and modules. The scheme helps analysts understand the model of a factory better and allows references to be made at a glance.
Clicking the items at the lowest level of the scheme, users open a section with accumulated information about each manufacturing unit: pages with a general overview of work and output, manufactured and defective products, materials used, losses, and labor. It is possible to set KPIs with different parameters (output, scrap, etc.) for each machine and use charts to track changes in absolute and relative terms. Analysts can check the efficiency of each machine: how many products were manufactured, at what speed, and with how many resources and losses.
Products, materials, and labor pages contain tables with more detailed information about each item. For instance, a labor page shows the actual number of workers vs standard brigade size, actual working hours and overtime vs standard shift time, and so on. To examine expenditures, users can look into tables with total salary and its breakdown by employees, working hours, and overtimes. As a result, such a structure enables analysts to identify bottlenecks and new growth points from the outset, and dig deeper into each parameter to find out the ways for improvement of the production process and financial performance.
Another part of our solution was the cost module. Its main page looks like a dashboard with a summary of the production costs of a chosen time period. It provides accumulated data about output, total costs, costs per unit produced, etc. Users can also compare data with a reference period to track changes over time. A small P&L table gives more information on spent cost per element (labor, raw materials, energy, depreciation, etc.) in comparison to the allocated budget and percentage of losses. For clarity, all cost categories are also presented in a graph. This overview allows users to understand at a glance the weight of each category in total production costs and identify problem areas for their optimization.
To get more details about cost structure, users can plunge into the information presented in large P&L spreadsheets, which are another part of the cost module. They include more parameters of each P&L component and allow users to calculate how production costs depend on output volume, price changes over years, and losses. The spreadsheet configuration already provides a list of general elements which are common for different types of plants – direct and indirect labour, raw materials, packaging, energy, maintenance, quality control, operating expenses, etc. Therefore, users only need to choose required elements in settings and adapt them to their own needs: they can change names if required, insert additional lines with components, and add actual costs and prices. The spreadsheet interface is especially easy to use: each row has a few buttons for editing data, adding a new nested row, and repeating and deleting the same row. All formulas and complicated calculations with many interrelated parameters are “hidden” in the backend, so ordinary users without admin rights won’t accidentally change or spoil them. The data can be imported from other tables (e.g. CSV, XML Spreadsheets), saving users a lot of time and minimizing manual work.
The next part of the cost module makes it possible to simulate top-down and bottom-up scenarios. Analysts can set reduction targets (e.g. by costs, losses) and get estimations of cost optimization by year. This helps to find the perfect combination of all parameters, allowing production companies to minimize costs at the same quality and increase their EBITDA. When a scenario is approved, it is used as a benchmark to track the current efficiency of production. Thus, analysts can detect serious errors in the efficacy and timing and recommend on taking corrective actions. In order to make the analysis of data faster and more efficient, our development team has implemented a Machine Learning algorithms module. Its aim is to define the patterns of changes in plants’ work parameters and analyze their effectiveness. These algorithms automatically receive the information from the ERP system and learn which ways were the most efficient in achieving the goals set for each plant, workshop or even machine. This allows the module to find trends and patterns that cannot be seen so easily but can be efficiently used in manufacturing cost optimization. This module works within the manufacturer’s perimeter and doesn’t label or store data about any branch, plant, workshop, or machine. Only statistical data is analyzed by the ML algorithms. This means that this module is secure and can’t provide the potential intruder or insider with any vital information on the manufacturing powers and items of the client.

PROCESS

The client had a clear vision of what the ideal analytical module should look like, so we got mock-ups of the interface and formulas to calculate the production costs. We solved the issues with everything else – software architecture, selection of the most suitable technologies and frameworks for flawless operation, frontend and backend development. Due to close cooperation of our Business Analyst with a strong financial background and the client’s analysts, together we managed to turn the complex logic of interrelations between all parameters into an efficient system of calculations on the Backend side and illustrative charts and tables on the user interface.


During the Discovery phase, our team examined the existing ERP platform of the client, dived deeply into the business processes, explored alternatives and figured out the best possible technological solution. We designed a model which outlines business entities in the system and their relationships, created a detailed plan for the development of each module and agreed with the client on the workflow.


We recommended technologies and frameworks that perfectly suited the client’s needs, according to their specifications and requirements. For instance, ngx-datatable was chosen to handle complex datasets in the format of spreadsheets. It provides all the necessary features to work with a table (sort, filter, add sub-rows, etc.) and is very flexible and light. D3.js was used for data visualizations and creating responsive chart components based on dynamic data. Thus, all the information was presented in an easy and illustrative form that enabled users to understand the data and make conclusions much faster. For Machine Learning we decided to use TensorFlow.js due to its high scalability and the possibility to use its capabilities on various devices in the future if needed.


After agreeing on the work breakdown structure with the client, we got down to developing the MVP. We delivered module by module every two weeks, discussed tasks in Slack and had weekly videoconferences in Google Meets. Each development phase concluded with unit and manual testing, so we could detect and fix even the smallest bugs at the earliest possible stage, preventing them from turning into issues.

PROJECT TEAM

PROJECT DURATION

TECHNOLOGIEN

OUTCOME

As the system was designed from the outset to be as user-friendly as possible it took just an hour to train employees working within a new module. Results became visible within a week of implementation: the preparation of reports became 3 times faster than before, and users got more time for closer examination of data and scenario simulation. This enabled analysts to come up with a more calibrated set of findings and recommendations for each factory within a shorter period of time, thus enhancing business value and increasing customer loyalty. Moreover, our team has managed to keep the whole ERP system as secure as it used to be before modifications were made thanks to the machine learning algorithms not interacting with any vital data while analyzing only patterns and trends that come out of certain decisions and actions.

Im Falle einer bestimmten Anfrage oder Technologie, die hier nicht aufgeführt ist, können Sie das mit einem Experten unter contact@innowise-group.com or Skype Innowise

SIE BRAUCHEN EINE TECHNISCHE LÖSUNG?

kontaktiere uns