Retrofitting or Remanufacturing Existing Assets with AI/ML Solutions

By Ebele Maduekwe

ARC Report Abstract


Machine learning (ML), artificial intelligence (AI), and other data science applications are proliferating across industry and infrastructure. Digitalization in general and Industrial IoT in particular have enabled massive quantities of data to be collected from connected machines to support machine learning to help improve asset availability and performance. Changing demographics and the need to increase workforce productivity and reduce costs are driving adoption of ML and other AI solutions.

However, this comes with high implementation costs. Also, a lack of appropriate technical know-how makes it challenging for industrial organizations to ride the digitalization wave to achieve their business goals.  This is particularly challenging for the many small-to-medium enterprises (SMEs) without ready access to data scientists to work with their subject matter experts and/or that lack the capital (or desire) to acquire new AI/ML-enabled assets.

In this report, we will show that existing “digitally challenged” assets can be retrofitted or, if needed, remanufactured cost effectively to support many digitalization goals. Using a simple net present value (NPV) approach, we also demonstrate how SMEs can assess the value of retrofitting or remanufacturing existing assets with ML/AI or other data science applications.

ML/AI for All

The convergence of information technology (IT), data science, and operational technology (OT) is driving the development of ML-enabled solutions for industry and infrastructure.  As applications in the IT domain move into the OT sphere, ML/AI or data science applications are increasingly needed, especially for deriving quick insights from ma-chine data.  Many large automation suppliers, machine builders, and end users are already incorporating digitalization into their respective business models, but participation among SMEs remains low.  

Unforeseen downtime, poor asset utilization, and quality control are just some of the challenges faced by today’s machine builders and machinery end users alike.  ML/AI solutions could help, but are often deemed economically impractical by SMEs.  However, investing in a smart, connected, AI/ML-enabled machine doesn’t necessarily require a costly complete overhaul, but rather small incremental changes to machine automation.

Dealing with Current "Digitally Challenged" Assets

Machines used in manufacturing are typically complex and expensive.  Machines used in metalwork, for example, require extensive recalibration and could be expensive to achieve.  SMEs use reamers, lathes, milling machines, and boring and shaping machines to produce precision parts for automotive and other industries. Due to the knowledge-intensive nature of many of these machines, their operation and maintenance are heavily affected by changing demographics and workforce. Even though most manually operated and basic numerical controlled (NC) machines have already been replaced by sophisticated computer numerical control (CNC) machines (many with Ethernet networking capabilities), few today come equipped with AI/ML capabilities.  How can SMEs decide on acquiring these capabilities on their “digitally challenged” assets?

Retrofitting and Remanufacturing to the Rescue

As previously mentioned, staying ahead in manufacturing today means digitalization and connected machines. However, purchasing new AI/ML-enabled machinery is expensive and often not practical for SMEs. This is where retrofitting or remanufacturing existing  machinery can provide an ideal solution.  In addition to avoiding the high capital costs, SMEs can save money as the newly retrofitted or remanufactured machinery fits in the same space and preserves the technical requirements needed.  This can often avoid the need for costly and time-consuming recertification.

When considering retrofitting or remanufacturing a machine, SMEs should evaluate the value of the investment based on overall cost, break-even margin, and long-term strategic outlook.  In the next sections, we use a simple net present value (NPV) approach to assess the payback from retrofitting or remanufacturing a three-axis CNC machine with connected capabilities and ML/AI software solution.

Case Modelling: CNC Machinery

remanufacturing Finished%20Metal%20Parts%20from%20a%20CNC%20Machine.JPGMachine retrofits or remanufacturing can be done in several ways. For example, a comprehensive retrofit on a CNC machine would involve a new machine electrical system, new digital servo motors and drives, feedback sensors, and new wiring with documentation on electrical semantics troubleshooting and guidelines.  A remanufacture means a complete overhaul of the mechanical, geometrical, motion control technology and electrical components including operator training. SMEs may request features like built-in hard drive for on-board file storage or Ethernet connection that enables ML/AI solutions.

Embarking on such an investment requires an assessment of the long-term strategic outlook and how profitable the investment will be. Using NPV, we first assess and estimate cash in-flow after the investment. Considering that the cash out-flow will be deducted from the discounted sum of cash- inflow, a positive net present value proves that the investment is cost-effective.


NPV estimates the current value of an investment by discounting the sum of all the cashflows resulting from the project and can be written as:

NPV=-C0+ i=1TCi1+ri

Above, C0 represents the initial investment, which is a negative cashflow, while the second half of the formula represents the discounted sum of all cash-inflow until a certain time point T.

Retrofitting or Remanufacturing a CNC Machine

CNC machines are usually large, complex and expensive with lifecycles ranging from four to more than 20 years.  Old machines are not usually IIoT enabled, spare parts can be scarce or discontinued; and most specialized machinists cannot afford to have unplanned downtime or fail while fulfilling an order.

For this hypothetical analysis, we used a variety of primary and secondary sources to establish the cost and profits involved in owning a three-axis CNC machine as well as the benefit associated with retrofitting or remanufacturing such a machine with ML/AI capabilities.  On average, the cost of a three-axis CNC machine ranges from $200,000 to $800,000.  A five-axis machine can cost up to $2 million. Buying a new three-axis machine with ML/AI or analytics capabilities could increase these costs by 50 percent. These costs are high for SMEs to bear. At the same time, due to competition, the orders handled by each machine, billed as CNC machining per hour, are typically low ranging from $50- $70 per hour for a three-axis machine and up to $300 per hour for a five-axis machine. Retrofitting or remanufacturing costs also vary by size and machine type. However, experts agree that the total cost of retrofitting should be less than 50 percent of the cost of a new machine.

Let’s suppose an SME would like to buy a new three-axis CNC machine for $500,000 with a lifecycle of eight years. In addition, the SME would also like to add IIoT connectivity and a simple ML/AI capability, such as the ability to build predictive models using updated machine software. The revenue from the machine at baseline is $70 per machining hour and machine runtime per day is restricted to eight hours with a cumulative monthly runtime of 160 hours (five days a week).  At baseline, this machine will generate annual revenue of $134,000 and take 3.7 years to pay back the initial investment of $500,000.

Assuming the SME pays an additional 20 percent of the machine cost (20 percent of $500,000) for retrofitting or remanufacturing. This upgrade improves lean production, eliminates waste, and results in a 1.5 percent increase in revenue for the second year, plus incremental returns of 0.5 percent for the third to fifth years. In addition to the physical and mechanical upgrades, an SME may decide to upgrade the CAD/CAM software used to program the machine with ML/AI capabilities. Traditional CAD/CAM software can be subscription based or sold for a one-time license charge ranging from $5,000 to over $20,000. Others may include yearly maintenance and update charges as high as $4,000 per year. Most machines seeking connected IIoT solutions may contain a combination of software charges for licensing as well as yearly maintenance and update charges.

For this upgrade, the SME includes CAD/CAM software for a three-axis CNC machine priced at $18,500 with a yearly maintenance and update charge of $3,500 at baseline (single machine user). Suppose that the software is upgraded to include a ML/AI solution, e.g. to support predictive maintenance or minimize metal waste. As calculated above, we estimate the cost of ML/AI at maximum, to be 50 percent of the baseline licensing fee and assume yearly maintenance and update charges to be constant. Accounting for all the above, one arrives at an improved CAD/CAM software with total costs of about $42,000 from time 0 to 5 years. This cost makes up 8 percent of the original price. We then estimate the physical and mechanical upgrades as the rest of the 12 percent (remember: retrofitting or remanufacturing costs are 20 percent of $500,000) as $60,000, bringing the total costs to $102,000. In total, the SME’s investment in an IIoT-connected 3-axis CNC machine, capable of supporting a ML/AI solution is calculated as roughly $602,000 ($500,000 + $ 102,000).


ARC Advisory Group clients can view the complete report at ARC Client Portal 

If you would like to buy this report or obtain information about how to become a client, please Contact Us     

Keywords: Artificial Intelligence (AI), Machine Learning (ML), Asset Utilization, Net Present Value (NPV), Connected Machines, Retrofit, Remanufacturing, ARC Advsory Group.

Engage with ARC Advisory Group