iRasus Technologies is leveraging the power of data analytics to identify the barriers that hinder the growth and adoption of electric batteries to develop a single platform interface with tools to integrate standards and protocols into one seamless experience for the consumer. Read on as co-founders Arjun Sinha Roy and Anirudh Ramesh tell Yashasvini Razdan from Electronics For You how their battery analytics startup is achieving these goals…
Q. Could you give me a brief about your product?
A. iRasus is an electricity data analytics company. We specialize (specialise) in electrical energy storage for two applications: the mobility segment and the stationary battery energy storage segment. We have an enterprise SaaS offering, which is meant for manufacturers of batteries. It’s meant for this industry’s electric vehicle fleets and certain ancillary uses. Regarding the mobility segment, we work directly with IPPs (Independent Power Producers) in the stationary segment.
The product helps ensure the batteries or energy storage systems are safe, reliable, and affordable. It also helps avoid premature degradation and fires caused by overheating the battery. In terms of affordability, it helps people monetize (monetise) EV fleets better through interesting business models such as battery-as-a-service (BaaS), enabling second life, and applications like that.
We use the data generated from the sensors inside the battery through IoT and analytics layers, feeding back into these business processes.
Q. How did you come up with the idea to make this?
A. Aniruddh and I are graduates of BITS Pilani but from different generations. We got in touch through the alumni network, and Anirudh was already working in the solar domain while I was in the telecom domain. We both saw that the next big leap in technology will be around the problem of climate change. I had done much work in telematics and IoT, and Anirudh specialised in renewable energy and its whole engineering aspect. So we decided to join hands and start this together. Anirudh is our co-founder and CTO, and he leads the engineering part. When we started, we were initially looking at the solar domain. Still, we quickly changed tracks towards storage because we felt that any transition from fossil fuel to clean energy would be driven by battery storage. Understanding and analysing battery storage and its data is crucial for successfully implementing this transition.
Q. Who is the target decision maker and the user of your product?
A. We started with battery manufacturers because they were facing the most problems with safety and reliability. Battery manufacturers are responsible for ensuring that the product they sell is safe when they sell it to an EV fleet owner. We’ve seen and read about incidents of battery fires and battery overheating sometime last year, and all these were ultimately attributable to the manufacturer. So the battery manufacturers were our first target market segment.
We gathered that the data has use cases for the fleet owners and fleet operators, and they became an equally important segment that we started focusing on. There were also others, such as people who set up and own chargers, who were a very important part of the puzzle. So we gave them the same software products associated with battery reliability and safety.
Q. What is the special innovation that you are trying to highlight?
A. We are looking at the utilisation of connected batteries and IoT sensor data to monetise EVs. The barriers to monetisation come in the form of three things — safety, reliability, and affordability. Software and data from these IoT sensors play an important role. We’re trying to gather all of this data and then make sense of it to ensure that the adoption of this technology increases. There’s also the element of lowering carbon emissions and moving towards sustainability. We believe there is going to be a huge challenge in integrating all these different infrastructure pieces and protocols, and their standards, into one seamless experience, in the clean technology sector. A single platform interface with a lot of these tools to integrate these different disparate standards, and protocols into one seamless experience for the consumer is what is needed and that is what we are building.
Q. What kind of data do you collect and analyse to provide insights?
A. There are two categories of data. We gather the data from sensors present inside the batteries that give you voltage, the current state of charge, the state of health, and a couple of other important indicators of status. We also collect geographical data, which includes latitude, longitude, speed, and direction. A combination of these will uniquely help you identify what the battery is doing and how it has performed in the past. The combination of past and present data gives you a lot of useful information on what to do with your battery
Q. Can you tell me more about your analytics process, from collecting data to generating insights and recommendations?
A. We look at different problem statements for analytics, trying to understand what the analytics has to do. Often, it is automating something that humans would already be doing in a certain context. If we talk about battery degradation, we look at the data and see the patterns that indicate some of the contributors to this degradation, such as an imbalance between the voltages of the cells.
The battery comprises multiple cells, each of which has its voltage. They all have to be in sync with each other. You cannot have one cell that is having a slightly different voltage compared to another cell. We built certain models on top of existing cell sell models that were already in place, that help you understand why some of these imbalances are occurring and ultimately predict the effect of this imbalance on whether degradation is going to occur. We took raw voltage measurements and plotted them over time.
We factored in things like certain states of the MOSFET inside the battery when the charge mounts a discharge, and its chemistry, and we arrived at a probabilistic estimate that the imbalance has been for x hours out of the total hours that the battery has been working. It lets us know the percentage chance we have within a stipulated number of days that some fault or error is going to happen. This is called exceedance probability and it becomes useful in understanding degradation. Similarly, certain models give you an exceedance probability of other events that could occur.
Q. How do you ensure the accuracy and reliability of your data analytics?
A. We ensure the completeness and accuracy of the measurements by collecting data from sensors inside the battery. These sensors are either part of the Battery Management System (BMS) or the telematics unit. Temperature sensors, integral to the battery, also contribute to accurate measurements. During production, measures are taken to ensure that the data collected by the sensors is reliable. To maintain data completeness, we capture measurements periodically using the telematics unit, ensuring minimal gaps in data reporting.
Every battery we work with is connected, capable of collecting sensor measurements and transmitting them via a telecom link using a 2G/3G/4G transceiver embedded in the telematics hardware. This serves as a gateway for transmitting both electrical and geographical data to the backend. Once the data arrives, we apply filtering to eliminate noise inherent in sensor data. We then transform the data into a format suitable for analytics, catering to various business cases such as warranty management and non-technical cases like battery theft. The telematics hardware acts as the data acquisition unit, guaranteeing both completeness and accuracy.
Q. The EV sector in India is still at a very early stage. Being an analytics startup, do you have enough data to process at this stage?
A. Heuristic models, which require substantial data for training and generating outcomes, are commonly used in machine learning models. However, we employ more analytical and heuristic models that rely on established relationships between inputs and outputs based on Fourier series or Fourier transform. These models do not require extensive training like traditional machine learning models. While we need some data, it is not as demanding as what is required for complex algorithms like ChatGPT, which require a large volume of data to produce meaningful results.
In our case, we require many outliers or anomalies to derive meaningful insights. For example, if provided with 1000 batteries, we would need at least 100 malfunctioning batteries to obtain valuable analysis. Obtaining such data can be challenging. Although it was more difficult when we started in 2019-2020, the industry has matured since then, and we now have access to a substantial amount of data highlighting issues in these batteries. This data facilitates modeling, albeit less complex than what is typically associated with algorithms like ChatGPT, which rely on vast amounts of data to deliver effective outputs.
Q. Can you give me some examples of how your analytics has helped companies in the EV sector make better decisions?
A. One of the initial use cases we focused on was warranty management. During discussions with battery manufacturers, we discovered that many products were being returned to the factory within six or seven months due to not meeting the desired battery lifespan. We identified the root cause as poor product design, with some components failing to meet appropriate quality standards.
Another category of issues we addressed involved problems arising from intermediate manufacturing steps in battery pack production, such as spot welding or wire bonding. Additionally, operational issues emerged, including improper use of the battery, such as harsh braking or using lead-acid battery chargers for lithium-ion batteries, which have different recharging curves.
Battery swapping also presented challenges. We encountered instances where faults were recorded in the swapped battery placed within the swapping station. Although the swap station initiated the charging process, the battery already exhibited a fault and should not have been charged as it could create faults elsewhere in the system. These operational challenges were identified through the availability of data.
Q. How do you deal with interoperability and integration with multiple technologies?
A. Interoperability poses a significant challenge in ensuring effective use of the software we deploy to our customers. Interoperability manifests in various forms. For instance, charging lithium-ion batteries with a lead-acid battery charger is prohibited due to safety concerns. Such integration of incompatible technologies necessitates the implementation of checks and balances within the software and firmware to prevent battery charging and potential instability.
Our software is integrated with numerous other software products that organizations may already employ. For example, we integrate data from battery manufacturers’ internal Enterprise Resource Planning (ERP) systems. Battery manufacturers can establish traceability by combining IoT sensor data with existing ERP data. This exemplifies the integration between two software technologies. While we continue to face integration challenges between different technologies, we are actively working on resolving them.
Q. How do you market this product and convince your target user to buy it?
A. We approach the market by identifying specific problem statements and presenting our solution accordingly. Compliance requirements, such as those outlined in AIS 156, an industry standard mandated by the government to ensure battery safety, are one aspect we address. Additionally, the subpar consumer experience in the EV sector emphasizes the need for data-driven consumer experiences. We pitch our platform to battery companies and Original Equipment Manufacturers (OEMs), showcasing the benefits and ROI (ROI) achievable through its implementation. This approach has been well received, as evidenced by recent contract signings. There is significant interest from battery manufacturers and OEMs in leveraging data-driven solutions. We fulfil an essential market need by prioritising software and data in conjunction with electronics and mechanical aspects.
Q. What are the advancements in battery analytics across the globe?
A. We have some global competitors who are doing analytics on battery data. I’ll tell you what are the different problems globally, which have been addressed using analytics. The first is doing things like simulations, and modelling of cells to design a better cell and subsequently a better pack. Some of the outcomes of these models have resulted in designing the million-mile battery that can help the vehicle run for at least 1 million miles before the battery dies. Some mature companies that are almost 10 years old, are looking at these problem statements for car makers such as Volkswagen, or Tesla in mature emerging markets like Northern America and Western Europe.
Q. How do you differentiate yourself from other analytics companies?
A. We are not contributing to designing cells or battery packs. We want to ensure that the right products are manufactured and used appropriately. There are two ways of analysing sensor data — you either look at individual cell voltages and multiply by the number of cells inside of the battery pack to define the behaviour of the battery. We follow a top-down approach where we look at how the battery as a whole is behaving and then try to break it down into its components, reducing the error that occurs in the first method.
Q. How are you ensuring data privacy and security, because this is all sensitive information that you’re collecting?
A. The luxury that we have is non-personal data. As a policy, we’ve maintained that all of the data is owned by the person who owns the battery. They are the beneficial owners of this data, we do not own the data. All that we do is make the software that helps in processing this data and then get something meaningful and useful out of it. We have read-only access to all of this data. What we’ve done internally is ensure that access control and restrictions for our use of data remain in place and then give our clients super admin access.
We have seen instances where our clients need to share data with other people with whom they work. We’ve built the software in such a way that they can decide what privacy settings and access control settings they want, in addition to doing periodic security checks, with whatever procedures they are doing. This framework has been working quite well so far because a lot of IoT companies manage their client data on their instances — whether it’s servers or any other IT hardware.
Q. What is your business strategy right now?
A. Business strategy is to now start selling in the market, we’re putting together a sales go-to-market strategy. We have already started selling and will now scale up our distribution as well. We are making some strategic partnerships in the industry and we will make those announcements shortly. We believe the market is taking off and there is a lot of demand for something like this. Our strategy is to increase our product offering, make it relevant to the market and start selling and scaling.
Q. Do you have any external partners who are helping you with this?
A. We are talking with a few partners in the semiconductor space. We are talking to one company which has a very strong presence in the battery management system sector (BMS). We are working with a few partners but a lot of these things are in the process of getting finalised and we are still in the early days.
Q. Are your partners based in India or outside of India?
A. As of now we are looking at the Indian market.
Q. What kind of challenges did you face in this journey?
A. I think, the only challenge that has always existed is customer education because it’s a nascent industry. I think that remains the biggest challenge. The need for data in a new industry takes a while to establish. We spend a lot of time and effort in explaining the value proposition to clients. We are seeing the fruits of that labour now.
Q. Have you started commercialising your product?
A. We started last year and have started commercialising our product from late last year and its early days of our revenue journey
Q. Have you sought any funding?
A. We raised $150,000 last year in an angel round.
Q. What is the cost of your product?
A. It’s a SaaS product function. The licence cost is a function of our energy modelling for different use cases. So, it’s approximately, the formula that we use is a percentage of the total energy value which goes in. Hence, it varies from industry to industry, sector to sector, so electric buses will be far more than the licence costs for two-wheelers and three-wheelers. We work with individual clients to understand the use case and then determine the pricing strategy, but it is a SaaS licence or battery per annum.
Q. How many people do you currently employ?
A. We have about ten engineers, and our total strength is about 15.
Q. Are you hiring right now?
A. We are hiring selectively for engineering roles.
Q. Do you hire freshers? Do you provide any internships? What are your hiring trends?
A. Yeah, we do have tie-ups with a few educational institutes. We have interns from NIT and BITS Pilani. We also hire people with experience.
Q. Do you face any challenges in training them or is it very seamless?
A. We faced quite a bit of a challenge earlier regarding talent acquisition, primarily because our space is at the confluence of different domains — electrical engineering, software product development, and analytics. We are now able to build a process where new talent in each of these domains can contribute to product development or the delivery of a particular product to a client or post-delivery customer service.
Q. Did you make any mistakes while developing this product? Do you feel you could have done something better?
A. We acknowledge that, during our journey, we encountered some critical improvement points. In 2019, our initial focus was primarily on solar power plants and EV chargers, overlooking the potential of batteries. However, we swiftly recognized the significance of batteries and made the necessary shift in focus. Another significant challenge emerged during the COVID-19 pandemic, impeding the commercialization of our software. Consequently, we had to recalibrate our plans, resulting in some setbacks. Nonetheless, we emerged stronger from this period of experimentation and have learned valuable lessons that continue to guide us.
Q. What are your upcoming innovations or your next milestone?
A. When we started commercialization, we were only focused on the two and three-wheeler battery segment. Since that product has been commercialised, we have shifted focus in the product development team to other products that are quite important in the market today. We see a couple of major areas to invest our time and resources. Primarily, we are looking at electric buses. We see that transportation services including shared mobility, and public transport, are the way forward. So we are building a similar product offering for electric buses which is our current focus.
We are also looking at the original applications and solar plants. We’re in discussion with many companies in the energy and utilities space and the power industry looking to adopt battery energy storage systems to store excess electricity from solar plants and wind turbines. We are building a product offering which functions as an energy management or a power management system on top of the supervisory control and data acquisition (SCADA) system for renewable energy, which will be stored in batteries.
We are also looking at smaller applications such as stationary batteries for commercial and industrial use. We are also looking at innovative use cases for the data in battery financing or asset financing cases. Insurance companies require this data for providing debt, credit or insurance. So, the data from these sensors and these IoT stacks are a good fit for applications like that. Similarly, for the second life of a battery that is once the battery is four or five years old, it needs to be refurbished or recycled into certain other applications. The data collected can be used here as well.
