Real Time Analytics in Smart Grids - The Art of the Possible

The next generation, forward thinking utility companies will differentiate themselves from their competition when they create an infrastructure that enables them to obtain what I call as the 'two second advantage'. Such an infrastructure depends on taking the base components, i.e. Distribution Management System, Outage Management System, Energy Management System, Power Generation, Mobile Workforce Management System, to name a few, and making them adaptive to real time changes in end to end value chain. Such responsiveness to changes that can be made effectual in real time and on-demand is where I feel the utility companies will differentiate themselves.
Let me take a step by step example on how analytics, based on both information and rest and information in motion can be combined to provide a compelling value proposition:

1. Using historical data (data at rest) from the consumer premises through the Meter Data Management systems (smart meters, concentrators, etc), one can create predictive models on how each region, territory and other geo-spreads have load usage. Predictive models can also be developed on the usage profiles on an hourly (or any other granular time unit) basis at a per consumer level or any other collective higher level of conglomorate.
2. Using historical data from power generation units, develop predictive models on generation trends and forecasts from each generation units.
3. Using historical data, develop predictive models on geo-areas which are more likely to be outage prone. Based on such models, develop predictive models on how the Mobile Workforce Management system would distribute or locate responders and dispatches.
4. Using the historical data from Asset Management systems, develop predictive models on assets which will require maintenance at given times. Such models will develop the analytical basis of a Condition Based Maintenance (CBM) plan.

The above 1, 2, 3 & 4 are just a few examples of how predictive models can be used to develop optimum Distribution Management System (DMS), Demand Forecasting, CBM and other plans, depending on historical data.

Now, the above are all good first steps in empowering the Smart Grid system with predictive analytics. The differentiated capabilities for the next generation Smart Grid comes from the next level of evolution from the above scenario wherein real time streaming information (information in motion) is overlaid on the parametric predictive models (as above) and enable dynamic adaptation of the predictive models and their real time redeployment so that immediate, real time actions can be taken. Such immediate, real time actions will empower the utility company to gain a significant advantage over its competitors and enable a real "Smart" grid framework for the society. And there is technology today to make this happen.

Let me cite an example on how the real time stream computing can facilitate what I just described above: Real time streaming data from the consumer smart meters flowing through the adaptive analytical engine can detect intra day shifts in usage patterns at households. The streaming data from weather forecasts can also predict hotter/colder intra day climates which along with the streaming data from the concentrators can dynamically recalibrate the static predictive models (Read 1 - 4 above) to create real time intra day demand forecasts or usages. This dynamically recalibrated load usage predictions will be used as a dynamic input to the generation unit. The generation unit can, in real time, reroute power based on such real time consumer usage in intra days. And this is just a start! In a related scenario, the real time streaming feeds from weather reports, social networking data and other sources can be correlated with data from the "Alternate Source" generation units to get a real time intra-day profile of power generation. Such a real time information can be used to recalibrate the distribution models at the generation units to dynamically find pockets of regions where real time load levels differ from the predicted ones and the models can work in conjunction in real time to dynamically conserve, reroute power. And think about how such real time, sub-millisecond dynamism can assist energy trading! In another scenario, the real time streaming textual data coming from social sites that captures sentimental and attitudinal data can be used to get a real time, on the fly capture of customer sentiments which can be used to influence the Customer Insight and Support system so that proactive measures can be taken to safeguard brand image and drive higher customer satisfaction. And lastly, Advanced Condition Based Maintenance is a current reality by mashing real time data from asset sensors along with advanced business rules to extend the component life of the critical and costly assets.

The power of real time stream computing coupled with data at rest is where the real value of information analytics is brought to the fore. This requires not only a futuristic look through the technology lens but also requires the culture and attitude in the company to be responsive and adaptive to changes.