Quantisweb Process Optimization as a Tool to Build a Bridge over the Skills Gap
By William Blasius
“Voice of the Customer” Advisor to Quantisweb Technologies
A Quantisweb Technologies White Paper
There is a growing concern within the processing industries, especially the chemical and material sectors, that there is an unbridgeable gap between the skills of older and new entrants. A contributing factor to this bifurcation is the scarcity of middle tenure employees. This leaves the burden of training and mentoring new engineers, chemists and physicists on already overburdened long tenure employees. These are the same people who are the last holders of critical tribal knowledge on process conditions and formulation protocols. Quantisweb Technologies optimization software can be a bridging solution by providing a means to model processes using either data mining or minimal experimental designs to help reduce daily firefighting and free up time for training. Quantisweb driven agile process and product development puts numbers to what had historically been rules of thumbs, allowing new employees to fully participate in innovation.
The chemical and material formulating and processing industries; especially the rubber and plastics, pulp and paper, ink and coatings and fibers and textiles segments, have recognized the growing technical knowledge gap between their veteran employees and more recent hires (1, 2). More disturbing is the bimodal distribution of tenure in knowledge intensive occupations. Rather than having a normal, bell shaped distribution with the bulk of employees filling the middle and tapering off in numbers at the two ends, the two ends are making up the bulk with a scarce middle. These are industries where it takes a solid five years of experience to function competently and semi-independently. Someone who will be retiring in three years does not have the time or motivation to train someone for five years.
There are multiple root causes for this phenomenon. Some are based on demographics with baby boomers and millennials simply outnumbering those in-between. Others can be traced back to the explosive growth of computer and biologic technologies and their attractiveness for graduating high school seniors looking for a major that would develop into a career in an expanding industry (3, 4). Finally, many of the merger and acquisition activities of the 1980’s and 1990’s found their financial synergies in the dismissal of many low seniority employees, who would have been the middle of the bell curve today.
The short term problems generated by a lack of mentoring/training can be softened by hiring in consultants and sending newer employees to university or industry sponsored trainings. The one thing that outside help cannot do is to help disperse the accumulated tribal knowledge of long tenured employees. There are two ways of doing that. The first is by formal apprenticeships or informal mentoring; the second is by efficiently generating new shared, deep knowledge. If a firm has sufficient slack resources, an apprenticeship is a great way to learn how rules of thumb came to be and how to navigate around red tape. What it cannot do is to generate a model of the process nor does it lend itself to getting beyond the universal answer of “because we’ve always done it that way”. It also keeps people locked in a cycle of not having enough time to fix things correctly, having less undistracted time because things are constantly coming undone because they are never fixed correctly and finally, being rewarded for fire-fighting with Band-Aid fixes. Creating shared, deep knowledge has a parallel with LeanSigma in that the ultimate goal is optimization, with the significant difference being that created shared deep knowledge becomes a bridging platform for future growth and innovation rather than being a basis for headcount reductions which further exasperates the skills gap.
In the early 1990’s, software was developed through a Canadian joint venture, to help companies optimize complex business processes. This has evolved into Quantisweb Technologies, a Montreal based software and technology company. The algorithms developed for stochastic approximation optimization have been complimented with the additional components of modern decision theory, minimally dynamic design of experiments (mdDOE) and analytical hierarchy process (AHP) to create a software system that can be considered agile and adaptive. Quantisweb software can take up to 200 input variables to optimize for 100 outputs. Inclusion of all potential variables is critical to reducing experimental bias. In development projects, the mdDOE can be used with mixed process and formulation variables using the number of variables plus one (n+1) to quickly understand what variables have effects or strong interactions and which are marginal distractions. This combination of functionalities offers firms an opportunity to concurrently understand their legacy formulations and processes better while creating new products faster with multiple data generated models.
New product developments in the mature chemical and material processing industries are most often incremental, expanding off successful legacy products. The knowledge and skills that created the original platforms are eroding as boomers retire out of the industry. Millennials, while well educated, lack industry specific expertise and experience. Teaming boomers with millennials with a tool like Quantisweb is a strategy to bridge the generational skills gap. By taking the time to utilize the experimental discipline inherent in mdDOE, younger employees will gain data and behavioral law based insight into formulations and processes that otherwise would be restricted to “gut” or “instincts” of more experienced employees. A fortunate output of such a bridge is that it serves as a gut check and a safe way to test why “we’ve always done it that way” without appearing as second guessing or disrespectful.
Given the popularity of LeanSigma styled programs, with the promise of increased productivity, reduced costs and ultimately greater short term profits, it is obvious that companies are dissatisfied with their current operational efficiencies. One thing that most of these programs lack is a means to attack multiple problems at the same time, often due to software variable input and output limitations. This means that one team can be working on maximizing throughput while another team is working on minimizing rejects. This leads to teams potentially coming up with completely opposite solutions for their micro-optimizations. By taking a global view of optimization, Quantisweb can help keep the firm working towards the same, common goal. Quantisweb can be used in three modes for optimizing legacy formulations/processes. The first mode is using the mdDOE function to design efficient experiments to minimize disruption to production schedules while maximizing information. As previously noted, Quantisweb mdDOE is based on n+1 experiments versus a standard full factorial DOE design which requires 2n experiments. The second mode would be to use Quantisweb to data mine all the inputs and outputs that have been previously collected by the firm’s production data acquisition systems to find a behavioral law for the collected information. Quantisweb essentially treats every minute of data acquisition as an experiment for analysis. The third mode is a combination of data mining followed by mdDOE to search out a previously unanticipated optimum. Once again, by applying the discipline of data collection, analysis and model building with experienced employees in collaboration with newer employees, the experienced employees get to check all the assumptions that had informally evolved from rules of thumb to facts. Newer employees get to see how the formulation and process variables interact in a data based, immediate feedback system (5). By creating this inter-generation bridge, the firm ends up with solid models for their current products, optimized processes and formulations and a technical bench whose training has been jump started based on solid experimental evidence and analysis.
Another positive effect of using a transparent, data driven system like Quantisweb is that recent hires can be integrated into the development and optimization processes with more direction than micro-managing. Given that the tenure of first job employees is on the decline (6), it is important to make new hires feel that they are making a significant positive contribution. With optimized formulations and processes, experienced employees will also have “found time” that can be used to help new employees navigate the firm’s systems and teach other soft skills to set them up for success. If done well, new employees can move up instead of out.
The disciplined optimization of processes allows longer tenure employees to free themselves from the drain of daily firefighting. These are exactly the employees with the talent and market knowledge to drive disruptive innovation. With Quantisweb minimally dynamic design of experiments, radical innovations can be pursued in small chunks which allows for a quick pivot as market conditions (competitive products, new suppliers, regulatory requirements) change. A data driven experimental system used as a tool in radical innovation can help create a bridge as experienced and newer employees learn together while building innovation platform models.
In conclusion, a knowledge creation and optimization tool such as Quantisweb can help bridge the generational skills gap by helping turn tacit knowledge into data based models, by giving new employees an opportunity to make fast, positive contributions to new and legacy products and by increasing innovation opportunities for all.
- “Bridging the Skills Gap: Help Wanted, Skills Lacking: Why the Mismatch in Today’s Economy?” American Society for Training & Development, October 2012.
- “Concerns Grow over Workforce Retirements and Skills Gap”, Theresa Minton-Eversole, Society for Human Resource Management, April 9, 2012.
- “Occupational changes during the 20th century”, Ian Wyatt and Daniel Heckler, Monthly Labor Review, US Bureau of Labor Statistics, March 2006.
- “Bachelor’s degrees conferred by postsecondary institutions, by field of study: Selected years, 1970-71 through 2011-12”, US National Center for Education Statistics, Digest of Education Statistics 2013.
- “Training Millennials in the Workplace? Teach Them the Same Way You Reach Them”, Julie Veloz, ADWEEK, On-line March 30, 2015.
- “Employee Tenure 2016”, US Bureau of Labor Statistics September 22, 2016, USDL-16-1867