On the 16 January 2019, Interlate, was proud to announce its partnership with Newcrest Mining Limited to provide expert decision support and near real-time productivity improvement services to Newcrest’s operations in Australia and Papua New Guinea.

“We are pleased to announce our partnership with Interlate to deliver their near real-time monitoring solution for our operations. Remote support that leverages advanced data analytics is a key part of our operational improvement strategy — Interlate’s Sentinel^TM service is the solution we were looking for, enabling those improvements to be sustained in the future,” said Craig Jetson, Executive General Manager Lihir and Cadia.

“The ability to rapidly identify significant value-adding initiatives made the decision to partner with Interlate easy,” said Craig. “Given their experience in remote monitoring, their solution provides us a ‘safe on-ramp’ into this space, without closing the door on developing our own capabilities in this area internally in the future.”

“We are delighted to be assisting Newcrest,” David Meldrum, the CEO of Interlate said. “Working together, Newcrest and Interlate have already discovered significant productivity improvement opportunities in Newcrest’s operations, which could contribute millions of dollars in profit each year. This value will steadily increase as Newcrest realises further gains from the real-time productivity improvement services.”

Client Challenge

The mining and minerals processing industry is a very large consumer of water and managing water consumption is a major issue for mining operations. Not only the high consumption of water, but the costs associated with it. From a social responsibility stand point, a mining operations reputation from an environmental performance point of view within the community is also an important consideration. Our client wanted to significantly reduce its water consumption and costs, and improve its environmental performance by optimising the water recovery process.

Approach

The experience of one of our ex Operating Mine Managers, who was familiar with operations  consuming a lot more water than the township and the community connected with that area was utilised. Based on past experience, we weren’t confident that the parameters around the recycling circuit for process water, like flocculant addition or thickener underflow densities, were actually being optimised for maximum process water recovery. Working with the client in a consultative approach, we were able to perform an analysis that looked at historical data combined with real-time data. Simultaneously,  multi variant optimisation of all of the parameters associated with water recovery was performed.

Result

Taking this approach, we were able to help our client significantly reduce its water consumption and costs, and improved its environmental performance by optimising the water recovery process. This strengthened the organisation’s position in communications with the community – “yes, we are huge consumer of water, relatively speaking, but we are also the most efficient users of recycle water that can be. And we can say that with our hands on our hearts”. This was a great outcome.

The challenge in the industry

It is no secret that aging infrastructure presents a key challenge to the water industry. As these assets deteriorate, they pose serious risks to both water supply and water quality. In sewer system networks, the risk of blockages is a major public health concern and the impacts to the environment, the water quality and the assets themselves can prove costly both from a monetary perspective and reputational risk.

The vision

Imagine if a pump operator could identify and solve blockages in sewerage pumps before they even occur. This process would allow them to optimise maintenance cost by providing data driven insights to maintenance and engineering teams about what can be done before a blockage even occurs.  Having the data available would enable operators to make informed decisions, reducing overall maintenance costs and increasing equipment availability to ensure business continuity.

Customer problem

We recently worked with a client that has approximately 800 pumping stations and each of these stations have varying quantities and ratings of pump drive trains, individual equipment from numerous manufacturers, and asset operation across vast geographical location with maintenance performed on a time – based schedule. To complicate this further, these assets are at varying stages of operational life, with equipment replacement planned based on operating time as opposed to asset condition.

Proposed Solution

The client was looking for a data driven condition based, predictive maintenance strategy across across pump station assets in their sewer system network.

Challenges encountered

Working in partnership with the client, the first sewerage pumping station identified had limited available data which was not unexpected due to the known age of the assets and subsequent data maturity. These data limitations meant that traditional analytics were not possible.

Overcoming the obstacles

An innovative approach was taken to develop a digital twin of the station by leveraging engineering and commissioning data. The digital twin was compared to the available operating data to validate the operating behaviour which provided confidence that the model could effectively soft-sense information not available from on-line instruments. Utilising subject matter expertise (SME’s), operating behaviours were identified and categorised based on their general impact on asset health.

Subsequently, a proof of concept (PoC) was designed to assess the asset health of the pumps across three of their sewerage stations by combining Advanced Data Analytics with motor and Variable Speed Drive (VSD) expertise with the aim of developing a predictive model to support a condition-based maintenance strategy.

Determining success

The success factors of the PoC relate to the development of an algorithm and / or tools that will: detect abnormal operation of pump station drive trains (motor / pump / VSD); predict failure of the pump station drive trains.

Outcome of the PoC

Working in partnership as an extension of the client’s team, advanced data analytics were applied to propose a predictive model to support a condition-based maintenance strategy. However, in PoC studies such as this, it has further highlighted the need to focus on the digital maturity of the assets across the entire operations.  Increasing the digital maturity would provide more informative data to be obtained. Increasing data maturity would enable clients to move from a predictive maintenance-based strategy to a prescriptive maintenance strategy. The power of this would ultimately ensure a calculation of lifetime left for the pump station drive trains possible, bringing a higher level of efficiency to asset management decision making.

Business Challenge

When a processing plant is already achieving extremely high throughput and world-class availability, how can improvements be made without significant capital spending? A globally diversified mining company was interested in understanding what further optimisation was available in a large coal processing plant, without spending additional capital. The site in focus already was already achieving world-class availability and throughput. The challenge was to find value that was not discoverable through traditional business improvement approaches.

Approach

Combining the capability of Galaxy Engine™, a hyper-variable analytical platform that is capable of processing millions of data points across discrete operating scenarios simultaneously at great speed,  with subject matter expertise and site operating context, a ‘Yield Optimisation Tool’ was developed to help the operator to achieve a target yield outcome by providing them with the best set-points for a selected coal feed.

The ‘Real-Time Optimising Platform’ was an ensemble of machine learning algorithms that can automatically determine the optimal set-points for the plant based on live data from coal coming into the plant.

Value realised

It was demonstrated that more than 2% annualised increase in yield was possible by optimising the set-point parameters of the circuit based on incoming coal feed. This improvement was worth more than $74M on an annualised basis with little or no additional capital expenditure.

Business Challenge

The processing plant had three major types of mineralisation and was looking for a way to model the behaviour and optimise output.

Our client’s plant has three distinct types of mineralised zones that were presented to the processing plant. Each of these zones have distinct requirements for processing and varying impacts on recovery and throughput. At any time the processing plant will be running a mix of these three mineralisations. The impact of these zones is not linear across the feed ratios. A solution was needed to be able to predict the expected result from any given mineralisation ratio that the plant was likely to work in. This information would also allow blend optimisation based on the available stocks of feed material.

Approach

Interlate applied a differentiated suite of technology and analytical solutions that could be implemented quickly to materially improve productivity, with little or no additional capital expenditure.

Using existing plant data, Interlate working longside the site team, was able to create a user interface. When combined with subject matter expertise and the operating context, this user interface allows rapid identification of predicted operating conditions based on current feed conditions, and visual indication of the current versus predicted operation conditions.

Value realised

Interlate used analytics, visualisation and subject matter expertise to identify productivity opportunities. This has allowed rapid identification of incoming ore related issues, and identification of where troubleshooting resources need to be utilised.

Business Challenge

Our client had a high throughput plant with significant subject matter expertise constraints on site. They were focused on identifying productivity improvement opportunities with low or zero implementation costs. 

Approach

Engaging Interlate for a value assessment, advanced analytics of 12 months of historical data from across the whole value chain was performed. Site machine learning, subject matter expertise and unique licensed technology were leveraged. Combing a value assessment with services provided by our data visualisation consultants, enabled the creation of simple visual tools for site operators and processing engineers to use to improve yield under a wide variety of circumstances. This value assessment is conducted over an intensive time period and recommendations provided within a 4-6 week period. In addition, online decision support is provided to assist site team to implement and verify improvements.

Value realised

Whole plant was analysed and focus areas were prioritised based on value improvement potential and being practical (both implementable and sustainable). Analyses highlighted advanced analytical capability providing significant productivity improvement. An additional ~290kT p.a. of in-specification coal product over the data set that is not discernible using conventional data analysis methods was dientified and an additional ~7% production for Operations. Product yield improvement of $30M annualised per annum were identified.

Business Challenge

Our client, a large Iron Ore producer was undertaking several initiatives aimed at increasing shareholder value. Some of these initiatives were to be funded by operational cash-flow however the challenge was to improve operational revenue without a significant increase in capital or operational expenditure.

Approach

We were engaged to provide technical consulting and expert data analytics to identify further optimisation opportunities for our client that were not discoverable using traditional approaches to business improvement. Using a consultative approach, a differentiated suite of technology and analytical solutions were applied and a unified data model that integrated mining and plant data was developed.

Result

Providing a combination of mining, processing, and mathematical expertise, practical tools that the site operational team could use to capture improvements in value in the plant were developed. Utilising the power of Interlate’s API Library, an ensemble of bespoke machine learning algorithms was used to perform sophisticated analyses to predict outcomes. The predictive intelligence that came from the machine learning based analytics gave the site team actionable insights. This tool was also applied to the stockpile building process and was able to predict ore feed categories of stockpiles to an actionable level of confidence, even with data that it had not previously seen.

Value realised

These solutions were able to implemented quickly to materially improve productivity, with little or no additional capital expenditure.  The optimisation achievable by using this approach was calculated at 3% increase in ore throughput and 2% improvement in product yield equating to an annualised revenue improvement of $57M.

Business Challenge

With our client being in it’s final stages of ramp-up to full production of its gold processing plant, their technical team were constantly seeking to help the company’s mining operations increase their profitability and were challenged to find solutions that created sustained improvement without spending additional capital.

Approach

Focusing on areas that would have an immediate positive impact on plant productivity, advanced analytics, combined with technology and expertise where used to uncover hidden value. With a focus on quantifying the proportion of lost throughput opportunity associated with ball mill recirculating load, an opportunity was identified to improve milling circuit throughput via tighter power draw control.

Leveraging the Galaxy Engine™ multiple data sources were fused into an integrated data model. Then with added subject matter expertise, Interlate developed an interactive decision tree, which generated discrete operating states for further interrogation. Further analysis of these states in collaboration with our client’s operations team uncovered a set of power draw operating tactics for the site to use to reduce throughput constraints.

Value realised

In just a few weeks, a 2% improvement in plant throughput was identified resulting in an annual revenue improvement of $4.2M per annum for the client.

Business Challenge

Our clients reliability system is world class, particularly in terms of data collection and incremental adjustments to equipment to ensure that remaining life is maintained and understood. The system is relatively labour intensive and utilises reliability personnel to input data into various spreadsheets to predict remaining life. This has proven highly effective at detecting failures in time to take action in the next maintenance cycle. Quantifying the impact of operating past functional failure could also be undertaken easily, informing the management team and allowing information driven decisions to be made in the event of conflicting priorities. This may become particularly useful as the end of life of mine approaches, because in the final twelve months of operation the conflict between productivity and maintenance cost management will become more acute.

Approach

A feasibility work program was undertaken following a successful Proof of Concept phase.  The Feasibility phase focussed on improving throughput through the development of a range of enhanced control strategies, as well as deploying the Interlate Sentinel^TM and Guardian services in real time.  Sentinel^TM and Guardian were deployed plant-wide and the resulting productivity improvements reflect opportunities that were identified  in multiple plant areas. 

Value realised

The feasibility work program, demonstrated how the Sentinel^TM, Guardian and Tactical Projects programs could be implemented to provide advanced real time productivity support. Sentinel^TM operations identified numerous productivity improvement opportunities, and demonstrated a reasonable level of decision support effectiveness.  The top four decision support interactions were quantified and demonstrated that an annualised metal production improvement of $7M could be achieved by implementing the outcomes of the agreed decision support records.