How IoT disrupt the Milk Industry

IMG_20161217_160213

Amul disrupted the Milk industry in 1950’s and caused what we popularly called milk revolution, since then annual milk production and collection has be raising annually over 10%.

Technology made its ways to many sectors, but still milk collection has be archaic and needs a second revolution.

Milk is collected in various villages across India in collection centers, which in turn is transported from collection centers to Milk Processing units. IMG_20161217_160213

The most important thing is what happens in the Milk collection center at villages, rather than Milk Processing units, because the quality of the end milk depends on maintaining the quality of the milk at village level.

Many co-operatives now use Milk Chillers at collection centers, in a way that solves the problem, but not quite enough.

The problem with having just chiller is that, what is happening to these devices will not be know to the Milk Processing plants.

How much milk is coming today ? What is the temperature of the milk ? Quality of the milk, these parameters needs to be monitored constantly to improve the ROI of the milk.

This is where IoT can be a game changer.  We  installed our IoT devices iIMG_20161217_163306n few places at Collection centers and delivered a game changing insights, which improved the collection ROI by 25%.

We also found out that Chillers were not being cleaned optimally, which was major cause of quality degradation of the milk at collection points.

What was considered unknown and un-manageable, suddenly become know and predictable.  The milk co-operative was able to not only got the accurate milk collection by 7 am in the morning, but can also constantly monitor the milk temperature remotely across their chilling centers.

 

Power Sector is moving from Energy Monitoring to Energy IoT as consumers becoming producers

Delta_images

When the power plant was brought up in the earlier days it use to be of the size of several 100’s of Megawatts, built in remote location and electricity was transmitted over 100’s of miles to the point of consumption. Because the installation runs in 100’s of millions of dollars, if not billions power plant operator can manage a team of experts installing and monitoring the system. These monitoring systems like SCADA can run into half-a million dollar type of cost and man power to manage it. This is still ok if you have large power plant.

Delta_images

With the advent of Solar, individual homeowners can produce power as low as 1KW and sell it to grid or to their neighbors.  Even small plant like this one cannot have the luxury of sophisticated monitoring system, until IoT comes into picture.  With IoT the entire site operations can be automated. Once the data goes to cloud in IoT framework, Machine Learning and AI algorithms can make the prediction of energy generation, any problem in the site or managing the maintenance schedule autonomous without any human intervention.

Telling the future of energy with data

When Facebook shows you a list of friends you may know, Google letting you know an ETD to be on time for a meeting, and other e-commerce websites giving you recommendations on things to purchase, these are instances where machine learning is carried out on large volumes of data called Big Data.

According to Gartner, the big data market is worth over $250 billion and surely it is here to stay. Businesses of all sizes that deal with various applications have started to adopt these practices.

Companies are now focused on how to store and manage this voluminous data. How should we architect the business’ technology stack to gain value from Big Data in terms of HDFS, complex event processing, NoSQL and machine learning? Store data on prem or cloud?

By means of advanced analytics, and machine learning, companies tap into their insight-rich vein of experience and mine it to automatically discover and generate predictive models to take advantage of all the data they are capturing. Departing from the traditional style of looking into the past for insights, companies can now predict parameters that they want knowledge about.

The value of machine learning is in finding structures that we have never seen before and precisely modelling to assist in decision making.

At​ TTC, we are leveraging these to build intelligent models that can serve our customers recommendations about optimising their usage patterns and first hand information about dynamic pricing for compliant infrastructures. We are developing these models in the energy sector where machine learning is hyper critical

Data Is the New Dinosaur

Data is becoming one heck of a challenge to solve. Take a example of our Product ThingsHiFi- a 5KW Solar Grid-Tie UPS. This device send data every 30 seconds to ThingsCloud. On a daily basis we will have

(30/60)*60*24 = 2,880 insertions

Each data packet is around 6K byte, which makes

2880*6k= 17.2 mb/per day

For one month,

17.2 *30 = 516 mb/per device/per day

You can see where I’m going from here, even if we have a modest 1000 devices, then we will be looking around 500 gigs of data per month. Right now we are exploring different architectures to process this data.  Also, inserted  record has around 10 values.

So, per month per device it is

2880*10*30 = 864,000 values

Now, with 100 devices, it would be 86 million values, which needs to be queried. Suddenly from nowhere we need to work on big data …!!!