We build executable decision systems grounded in formal mathematical methods: Mixed-Integer Programming (MIP), Decision Diagrams (DDs), Network Flow Models, and Dynamic Programming.
These are not spreadsheets or simulation toys. They are industrial-grade optimization engines that deliver provably optimal or near-optimal solutions to complex problems where intuition fails and trial-and-error is too expensive.
A formally optimized system configuration ensures maximum business value under real-world constraints.
Clients face dozens of technically feasible designs for new facilities or revamps, but lack rigorous methods to select the best one.
Formulate as a Mixed-Integer Program (MIP) using CPLEX/Gurobi to optimize CAPEX, availability, maintenance burden, and EBITDA impact.
Optimized the design of a major LNG compression train by evaluating over 348 configurations under constraints on redundancy, feed variability, and crew logistics.
Delivered proof of 98.7% optimality; unlocked $MM in EBITDA through superior system configuration.
A robust maintenance schedule integrates operational rules, resource limits, and risk exposure into a single executable plan.
Large plants must coordinate hundreds of tasks across crews, spares, weather windows, and production commitments — a combinatorial challenge beyond manual planning.
Apply Relaxed Decision Diagrams (R-DDs) to compactly represent feasible schedules, enforce logical constraints, and provide relaxation bounds on optimal performance.
Developed an optimized maintenance schedule for a large gas processing plant, eliminating infeasible sequences upfront and reducing downtime risk by 30%.
Generated a conflict-free plan aligned with real-world constraints — not theoretical assumptions.
An optimized flow network maximizes throughput while minimizing cost, energy use, and risk across interconnected assets.
Gas pipelines, electrical microgrids, and steam systems suffer from inefficiencies due to suboptimal routing and load balancing.
Model as Minimum-Cost Flow Problems to minimize pumping/fuel costs while maximizing delivery reliability and redundancy utilization.
Optimized power routing between grid, solar, battery, and diesel generators at a hybrid energy site, reducing fuel consumption by 22%.
Turned intermittent supply into reliable, low-cost operations.
A decomposed optimization framework enables solution of enterprise-wide problems by coordinating subsystem-level decisions.
Multi-site asset fleets or enterprise-wide decarbonization programs exceed solver capacity when modeled monolithically.
Use Dantzig-Wolfe or Benders Decomposition to break problems into tractable subproblems solved iteratively, then recombine via master problem.
Applied decomposition to optimize maintenance scheduling across a national hydropower fleet under shared resource constraints.
Solved what was previously considered intractable — without oversimplification.
A multi-stage investment policy adapts to evolving risks and opportunities, preserving value across uncertain futures.
Should we delay revamping? Upgrade modularly? Or replace entirely? Most firms rely on deterministic projections.
Use Stochastic Dynamic Programming (DP) and Reinforcement Learning to simulate thousands of scenarios and prescribe adaptive investment strategies.
Evaluated phased expansion pathways for a chemicals manufacturer under volatile feedstock prices and regulatory risk.
Prescribed an adaptive policy with 95% confidence of superior ROI over static alternatives.
Unlike heuristics or manual planning, formal optimization delivers mathematical proof that you've found the best solution — or quantifies exactly how close you are to optimal.
Physical limits, operational rules, safety requirements, and contractual obligations are encoded as hard constraints — guaranteeing feasibility before deployment.
Formal methods handle problems with thousands of variables and constraints that overwhelm spreadsheet analysis or simulation-based trial-and-error.
Stochastic and robust optimization frameworks explicitly model uncertainty and prescribe adaptive strategies that preserve value across multiple scenarios.
This is engineering, not consulting. We co-architect systems where value is mathematically maximized, not guessed.
At Knar, we walk beside mature organizations through complexity. Our engagements are not about delivering answers, but about co-architecting systems where technical reality meets financial performance, data becomes institutional knowledge, and decisions are made with traceability, not intuition.
Let's explore what optimal looks like for your next project.
Contact Us